Accurate stratigraphic ages are crucial to understanding the deformation history of the Tibetan Plateau prior to and during the Indo-Asian collision. Efforts to quantify Mesozoic-Cenozoic ages are hindered by limited fossils and a paucity of volcanic horizons and regionally correlative strata. Magnetostratigraphic and biostratigraphic results from the Xining-Minhe-Longzhong basin complex and Dangchang basin provide an improved chronology of nonmarine basin development over a large region of the northeastern Tibetan Plateau (34–37°N, 101–105°E). Analyses of 171 magnetostratigraphic levels and 24 palynological assemblages (>120 species) indicate Late Jurassic-Early Cretaceous to mid-Tertiary deposition. Although magnetic polarity zonation is incomplete, independent palynological age control partially restricts possible correlations to the Geomagnetic Polarity Timescale. The sediment accumulation record, basin provenance, structural geology, and published thermochronological data support a history of Jurassic exhumation, Late Jurassic-Early Cretaceous fault-related basin initiation, and Cretaceous-Paleogene reduced accumulation. These patterns, which are compatible with Late Jurassic-Early Cretaceous extension and Cretaceous-Paleogene postrift thermal subsidence, were disrupted at about 40–30 Ma, when shortening related to the Indo-Asian collision induced localized range uplift, vertical axis rotation, and amplified subsidence.
 Obstacles to chronological studies in the Tibetan Plateau include a shortage of diagnostic fossils, pronounced lithostratigraphic similarities (precluding unique regional correlations), and limited volcanic horizons in Upper Jurassic–Tertiary strata [Kidd et al., 1988; Yin et al., 1988]. Although marine strata are reported in places [Zhang, 2000] and important mammal fossils characterize some Neogene deposits [Qiu et al., 2001; Wang et al., 2003], most mid-Mesozoic through Cenozoic sedimentary rocks consist of nonfossiliferous, oxidized red beds of fluviolacustrine origin. Because of the scarcity of published age constraints, several regional studies rely exclusively on age interpretations from regional Chinese mapping efforts [e.g., Halim et al., 1998; Métivier et al., 1998; Cogné et al., 1999]. In many cases, however, the rationale behind these age interpretations and regional correlations, as well as the original sample locations and stratigraphic levels, remain unspecified in the available literature. Furthermore, in some instances, dating of volcanic horizons reveals discrepancies with original map-assigned ages on the order of 20–80 Myr [e.g., Horton et al., 2002; Kapp et al., 2002].
 The purpose of this paper is to develop a chronological framework on the basis of magnetostratigraphy and palynological biostratigraphy to evaluate late Mesozoic to mid-Tertiary basin evolution and sediment accumulation patterns in the northeastern Tibetan Plateau (Figure 1). These chronostratigraphic techniques may provide the best clues to distinguishing Mesozoic rocks from lithologically similar Cenozoic rocks in other parts of the Tibetan plateau. In the companion paper, Dupont-Nivet et al.  address tectonic rotations revealed by the paleomagnetic analyses.
 A complex stratigraphic nomenclature characterizes Mesozoic-Cenozoic basins in the vicinity of Xining and Lanzhou, two cities near the Qinghai-Gansu provincial border (Figure 2). Most successions contain predominantly Cretaceous red clastic strata that persist laterally throughout the region and originally covered >30,000 km2. An Upper Jurassic section concordantly underlies these rocks in places. The Jurassic-Cretaceous package is referred to as the Xining-Minhe basin or Minhe basin. The approximate outcrop extent (Figure 2) suggests an original southern basin margin south of the Laji Shan, eastern basin margin near the Maxian Shan (near Lanzhou), northern basin margin near the Qilian Shan (Daban Shan, locally), and western basin margin near the Qinghai Nan Shan. Lower Tertiary rocks that concordantly overlie Cretaceous strata in most of the Xining-Minhe basin have been traditionally considered part of a younger broader basin, the Longzhong basin [Zhai and Cai, 1984]. Nomenclature is further complicated by the delineation of four Cenozoic subbasins (the Xining, Lanzhou, Linxia and Longxi-Jingning subbasins) within the composite Longzhong basin (Figure 2b). For consistency with previous studies, the original names are retained here and the term Xining-Minhe-Longzhong basin complex is introduced to refer jointly to the Mesozoic Xining-Minhe basin and Cenozoic Longzhong basin. The Longzhong basin covered >100,000 km2 between the Qinling Shan and Haiyuan fault during Cenozoic sedimentation (Figure 2). Paleogene strata are best preserved in the Xining subbasin and along the northern flank of the Qinling Shan. During Oligocene-Miocene time, deposition persisted in these areas and initiated in the Linxia and Lanzhou subbasins and a series of basins southwest of the Laji Shan: the Gonghe, Guide, and Xunhua basins (Figure 2b) [Zhai and Cai, 1984; Li, 1995; Li et al., 1997a, 1997b; Flynn et al., 1999; Qiu et al., 2001; Yue et al., 2001; Pares et al., 2003; Song et al., 2003]. Variability in basin nomenclature may reflect systematic partitioning of the original Mesozoic Xining-Minhe basin by Cenozoic uplift of narrow ranges.
3.2. Xining-Minhe-Longzhong Basin Complex and Dangchang Basin
 (1) Middle Jurassic Yaojie and Xiantang formations contain ostracod Darwinula and palynomorphs Cyathidites, Deltoidospora, Classopollis, and Osmundacidites [QBGMR, 1985; Pang and Whatley, 1990]. (2) Upper Jurassic Datonghe Formation contains ostracods Damonella, Djungarica, and Minheella, and charophytes Aclistochara and Latochara, similar to Late Jurassic assemblages of China, Mongolia, and western United States [Su et al., 1983; Hao, 1988; Li, 1988; Pang and Whatley, 1990]. (3) Lower Cretaceous Hekou Group contains ostracods Cypridea, Lycopterocypris, and Ziziphocypris, conchostracans Eosestheria and Yanjiestheria, charophytes Aclistochara and Mesochara, and palynomorphs Cicatricosisporites, Schizaeoisporites, Classopollis, Piceaepollenites, Deltoidospora, and Cedripites. The ostracods, charophytes, and conchostracans are common to abundant in Lower Cretaceous rocks of Eurasia and western United States [Su et al., 1983; Hao, 1988; Li, 1988; Pang and Whatley, 1990; QBGMR, 1991; Tang et al., 2001]. (4) Upper Cretaceous Minhe Group contains ostracods Cristocypridea, Talicypridea, Cypridea, and Candona, charophyte Latochara, and palynomorphs Schizaeoisporites, Classopollis, Ephedripites, and Tricolporopollenites. All are typical of Upper Cretaceous strata in China and Mongolia [Su et al., 1983; Zhai and Cai, 1984; Hao, 1988; QBGMR, 1991]. (5) Paleocene to possibly lower Eocene Qijiachuan Formation contains large quantities of ostracods Cypris and Limnocythere, charophytes Stephanochara, Harrisisichara, Peckichara, and Rhabdochara, and palynomorphs Ephedripites, Proteacidites, Scabiosapollenites, and Normapollis, similar to upper Paleocene and possibly lower Eocene strata in China [Zhai and Cai, 1984; Guan, 1988; Hao, 1988; QBGMR, 1991]. (6) The Eocene Honggou Formation contains ostracods Limnocythere, Ilyocypris, and Cyprinotus, charophyte Gyrogona (the main component of middle to upper Eocene strata in China), and palynomorphs Quercoidites, Celtis, Ulmipollenites, and Iodes [Zhai and Cai, 1984; Hao, 1988; He et al., 1988; QBGMR, 1991; Liu and Yang, 1999]. (7) Oligocene Mahalagou Formation contains a large appearance of ostracod Ilyocypris, additional ostracods Cyprinotus and Eucypris, charophytes Maedlersphaera and Sphaerochara, and palynomorphs Meliaceoidites, Piceapollenites, Ephedripites, and Chenopodipollis, common in Oligocene strata of Eurasia [Zhai and Cai, 1984; Hao, 1988; QBGMR, 1991; Yu et al., 2001]. (8) Miocene Xiejia, Chetougou, and Xianshuihe formations are characterized by the appearance of a tremendous range of vertebrate fossils [Zhai and Cai, 1984] and the presence of ostracods Eucypris and Ilyocypris, charophytes Charites, and palynomorphs Potamogeton, Quercoidites, and Gramineae [Zhai and Cai, 1984; Hao, 1988; QBGMR, 1991; Yu et al., 2001].
 In addition to fossil assemblages, Eocene-Oligocene marker horizons (gypsum beds in the Hongguo and Mahalagou formations) facilitate regional stratigraphic correlations. For example, a 5- to 30-m-thick gypsum in the upper Hongguo Formation can be traced >50 km, from local salt mines northwest of Xining to south of Pingan (Figure 2). This unit (E2h3) and similar units visible on aerial photograph and satellite imagery are located and described in previous maps and reports [Zhai and Cai, 1984; QBGMR, 1985].
3.3. Additional Palynological Age Control
 We present >120 palynological species identified in 24 samples from Jurassic-Tertiary strata (Tables A1 and A2), adding to the ∼80 previously reported species (Figure 4). For each sample, 5–10 grams of siltstone was dissolved in HCl and HF acids. Resistant pollen grains were sieved with 10 μm mesh, floated in a solution of ZnBr2 (specific gravity of 2.0), mounted on standard slides, and examined using a binocular microscope. An additional 17 samples were barren of palynomorphs. Jurassic-Cretaceous palynological assemblages are dominated by gymnosperms and pteridophytes (ferns), with small proportions of angiosperms in Upper Cretaceous strata. Tertiary assemblages are dominated by angiosperms with small but regular occurrences of gymnosperm and pteridophytic flora. Zhai and Cai  noted similar distributions in the Xining-Minhe basin.
3.4. Depositional Environments
 Jurassic-Tertiary strata are composed of lithofacies associations attributable to deposition in lacustrine and distal fluvial environments with subordinate proximal fluvial and alluvial fan environments. Comprehensive descriptions and interpretations, based on measured sections (Figure 5), are presented in Table A3.
 Lacustrine deposits, the volumetrically dominant facies, consist of thick intervals (100–1000 m) of thin-bedded, moderately well sorted, red, yellow, and variegated mudstone, sandstone, and minor evaporite. Deposition is attributed to suspension fallout and turbidity currents, with gypsum precipitation occurring during evaporative playa lake conditions, most notably during Eocene-Oligocene time.
 Overbank fluvial deposits are widespread, consisting of moderately thick intervals (20–200 m) of thin- to medium-bedded tan and red sandstone and interbedded mudstone. Clastic deposition is interpreted to be the result of crevasse splays and suspension fallout in an overbank fluvial setting, with minor overprinting by pedogenic processes.
 Proximal fluvial deposits, most common in the southern and eastern portion of the Xining-Minhe basin, compose stratigraphic intervals 10–200 m thick containing medium- to thick-bedded, moderate to poorly sorted, red and tan sandstone and sandy conglomerate. These deposits represent sedimentation by bars and dunes within erosive fluvial channels.
 Alluvial fan deposits, best developed in the Dangchang basin and southeastern Xining-Minhe basin, define stratigraphic intervals 20–400 m thick consisting of medium- to thick-bedded, poorly sorted, clast-supported red conglomerate and interbedded sandstone. These strata recorded gravel and sand deposition by fluid flood flows, hyperconcentrated flows, and rare debris flows.
4.1. Paleomagnetic Sampling and Analysis
 Sampling for magnetic polarity stratigraphy was carried out in the central Xining-Minhe basin (east Xining section: 36°35′N, 101°54′E) and Dangchang basin (north Dangchang section: 34°05′N, 104°28′E) (Figure 2). Regionally, these localities afford the most continuous exposure of suitable fine-grained lithologies, greatest palynological age control, and greatest structural integrity. Sampling was performed in situ using a portable, gasoline-powered drill. Two to four samples were collected per magnetostratigraphic level (each level representing a horizon <1.0 m thick) at an average stratigraphic spacing of ∼10 m. Red mudstone and sandstone were sampled preferentially where possible. A total of 171 magnetostratigraphic levels were collected: 93 Dangchang levels and 78 Xining levels. Magnetostratigraphic sampling was complemented by the regional paleomagnetic study of Dupont-Nivet et al. , wherein data tables and complete descriptions of paleomagnetic analyses and thermal demagnetization techniques are provided.
 At Dangchang, 85 magnetostratigraphic levels (Figure 5) were collected within a continuous, ∼1100-m-thick, southwest dipping homoclinal succession. To perform a fold test, 8 additional paleomagnetic sites (8 samples per site) were collected within strata that are gently folded on a wavelength of several hundred meters and display shallow to moderate (15–34°) westward and northward dips. Upon stepwise thermal demagnetization (∼20 temperature steps) of one pilot sample per collected level, 80 of 85 levels yielded interpretable characteristic remanent magnetization (ChRM) directions. Samples from the other five levels, and two additional levels exhibiting aberrant ChRM directions, were discarded from further analysis. Demagnetization behavior is consistent with magnetizations carried by a combination of magnetite and hematite [Dupont-Nivet et al., 2004]. Because normal polarities were obtained from all analyzed levels, no further demagnetization experiments were performed. Therefore each of the calculated virtual geomagnetic pole (VGP) latitudes for the Dangchang section (Figure 5) represents the ChRM direction of a single sample. Positive VGP latitudes were calculated for all levels: the majority are in the 40–80° range, but two (775 and 1025 m levels) exhibit anomalously low values (<25°). The results from these two levels, although suggestive of aberrant readings or local complexities, do not change the interpretation of a single normal polarity zone for the Dangchang section. From the 8 paleomagnetic sites collected to perform the fold test, 8 samples per site were thermally demagnetized and ChRM directions were defined. Site-mean directions cluster after structural correction, indicating passage of the fold test [McFadden, 1990] and a prefolding origin for the ChRM [Dupont-Nivet et al., 2004]. A lack of reverse polarities throughout the section precludes application of the reversals test.
 The Xining data set (Figure 5) was collected from a continuous, ∼800-m-thick, north dipping homoclinal succession that exhibits sufficient variations in bedding attitude suitable for a fold test. Sampling was divided into lower, intermediate, and upper stratigraphic levels. For the lower section (12–356 m), 55 magnetostratigraphic levels were collected at an average stratigraphic interval of ∼6 m. Within intermediate stratigraphic levels (356–553 m), very poorly indurated mudstone and interbedded gypsum prevented sampling. For the upper section (553–801 m), 23 magnetostratigraphic levels were collected at an average stratigraphic interval of ∼11 m. For the lower part of the Xining section, 49 of 55 levels yielded interpretable ChRM directions carried mainly by hematite. Three aberrant ChRM directions were discarded. For the remaining 46 levels, ChRM directions pass both the reversals test and the fold test [McFadden, 1990; McFadden and McElhinny, 1990], consistent with acquisition of the ChRM during or shortly after deposition. For the upper part of the Xining section, 22 of 23 levels yielded ChRM directions carried mainly by magnetite. Although variations in bedding attitudes are insufficient to allow a determinate fold test, the ChRM directions pass the reversals test [McFadden and McElhinny, 1990], suggestive of a primary origin for the magnetization.
4.2. Age Correlation
 A limited quantity of magnetic polarity zones are defined for the Dangchang and Xining sections, precluding rigorous correlation of polarity zonation to the geomagnetic polarity timescale (GPTS) on the basis of pattern recognition. However, the fossil age data (Figure 4 and Tables A1 and A2) partially restrict possible correlations. Our rationale therefore is to utilize the combination of fossil ages and polarity zonation to define a single reasonable correlation for each section, rather than to match a series of alternative correlations to the GPTS. It must be emphasized that each correlation shown in Figure 5 represents only one of several possible correlations.
 For the Dangchang basin, exclusively normal polarities are observed for all 78 magnetostratigraphic levels (350–1437 m level). The uniformity of magnetic polarity, large number of analyzed levels, and ∼10 m sampling interval suggest that it is unlikely that significant zones of reversed polarity were missed within the 1087 m sampled interval. Age constraints from new palynological assemblages (433, 596, 1306, and 1341 m levels) consistently indicate an Aptian-Albian (late Early Cretaceous) age (Table A2). Therefore the single, 1087-m-thick, normal polarity zone is correlated to part of the Cretaceous normal polarity superchron (chron 34n). The combination of exclusively normal polarities and Aptian-Albian palynoflora further restrict deposition to a time interval ranging from 120 Ma (base of chron 34n) to 99 Ma (end of Albian), based on the GPTS of Cande and Kent  and Gradstein et al. . This 21 Myr window represents the maximum duration for deposition; the possibility of deposition over a shorter time frame cannot be ruled out.
 Sampling of the Xining section was divided into lower, intermediate, and upper stratigraphic levels. For lower levels (12–356 m), 36 levels yield normal polarity ChRM directions and reversed polarities are observed at 10 levels. Two normal and three reverse polarity zones are each defined by two or more levels of similar paleomagnetic polarity; a third normal polarity zone at the top of the sampled interval is inferred on the basis of a single level. Eight palynological assemblages (12, 13, 17, 27, 247, 248, 293, and 351 m levels) indicate a Late Jurassic or Early Cretaceous age for the lower Xining section (Table A1). There are many possible correlations of the six polarity zones to the GPTS. In the correlation presented, the Aptian-Cenomanian (121–93 Ma) palynological assemblage at the 351 m level is assumed to lie within the Cretaceous normal polarity superchron (chron 34n of Cande and Kent  and Gradstein et al. ), which spans from early Aptian (120 Ma) through Santonian (83 Ma). Therefore the stratigraphically highest normal polarity zone, defined by a magnetostratigraphic level at 356 m, is correlated with the lowermost part of chron 34n. Further correlations of underlying polarity zones are as follows: a narrow reversed zone (306–330 m) to chron M0r, a thick normal zone (214–306 m) to chrons M1–M3n, a reversed zone (178–214 m) to chron M3r, a thick normal zone (22–178 m) to chrons M5n–M11n, and the basal reversed zone (12–22 m) to chron M11r of Gradstein et al. . The correlation presented provides a reasonable explanation for the series of palynological assemblages from the lower 400 m of the Xining section, but alternative correlations are possible.
 Although paleomagnetic data are not available for intermediate levels (356–569 m) of the Xining section, new palynological assemblages indicate an Aptian-Cenomanian (mid-Cretaceous) age at the 351 m level and a Cenomanian-Maastrichtian (Late Cretaceous) age at the 535 m level (Table A1).
 For upper stratigraphic levels (569–801 m) of the Xining section, 22 magnetostratigraphic levels (10 normal polarity, 12 reverse polarity) can be grouped into six polarity zones. Two normal and three reverse polarity zones are each defined by two or more levels of similar polarity; a third normal polarity zone at the top of the sampled interval is inferred on the basis of a single level. However, the wide spacing between magnetostratigraphic levels (a result of limited suitable lithologies) virtually ensures that the observed magnetic polarity zonation will incompletely identify reversals of the GPTS. Utilizing gypsum marker units, new chronostratigraphic data from Paleocene and Eocene palynological assemblages (Table A1) collected nearby at the Pingan section are correlated along strike to the Xining section at the 553, 559, 566, 582, and 865 m levels (Figure 5). A tentative correlation to the GPTS is made wherein the lowest magnetostratigraphic level (569 m) is considered to be near the Cretaceous-Paleocene boundary, based on Late Cretaceous palynoflora at 535 m and Paleocene-Eocene palynoflora at 553–582 m. Further correlations of overlying polarity zones are as follows: a reverse polarity zone (569–604 m) to Paleocene chrons 26r–27r, a normal polarity zone (661–729 m) to early Eocene chrons 21n–24n, and a reverse polarity zone (729–795 m) to middle Eocene reverse chrons 18r–20r. Although offering a reasonable explanation for palynological assemblages, the magnetic polarity correlation presented for the upper Xining section is nonunique. These results are attributable to incomplete sampling of the numerous polarity chrons that characterize the GPTS from Late Cretaceous to Oligocene time.
4.3. Accumulation Rates
 Sediment accumulation rates are calculated on the basis of magnetic polarity correlations shown in Figure 5. All rates discussed here are subject to a host of uncertainties, including the errors in magnetic correlations (mentioned previously), errors in the GPTS, resolution of palynological assemblages, uncertainties related to unsteady deposition and stratigraphic completeness, errors in measured stratal thicknesses, and postdepositional thickness changes [May et al., 1985; Badgley et al., 1986; Gallagher, 1989; McRae, 1990]. One important issue concerns the possibility of significant time gaps within the stratigraphic successions. Although angular unconformities are not observed, unsteady deposition could result in disconformities of several million years duration (e.g., Jurassic section of Qaidam basin [Ritts and Biffi, 2000]). For the Xining succession, the potential for unconformable time gaps is minimized by several factors. First, the upper and lower age bounds indicate a 107–108 year duration of potential deposition, exceeding the 100–105 year timescales over which accumulation rates are most sensitive to the effects of unsteady, discontinuous sedimentation [Sadler, 1981; May et al., 1985; Badgley et al., 1986; McRae, 1990]. Second, the volumetric dominance and large vertical extent of lacustrine facies (Table A3) suggests that (1) erosion of previously deposited sediment, a process common in channelized fluvial systems, was of reduced importance, and (2) deposition was steadier than in a river-dominated system, as demonstrated by Sadler . Third, the paucity of frequent or mature paleosols in overbank (nonchannelized) fluvial deposits suggests, qualitatively and possibly quantitatively [Kraus and Bown, 1993; Retallack, 1998], that there were not prolonged depositional hiatuses. Additional uncertainties to consider include postdepositional variations in stratal thickness, particularly the effects of compaction. The long-term average rates reported in Figure 5 represent compacted stratigraphic thickness per elapsed time. To remove the effects of compaction beneath younger sedimentary overburdens, a standard backstripping analysis [Van Hinte, 1978] was performed assuming reasonable porosity values [Sclater and Christie, 1980; Angevine et al., 1990], yielding decompacted rates about 10–30% higher. Given all the possible sources of error, the reported long-term average rates are considered approximations of actual accumulation rates. Significant variations in the actual rates over time frames <106–107 years may remain undetected by the analysis presented here.
 For the Dangchang basin, a minimum average rate of 5.0 cm/kyr or 50 m/Myr (55 m/Myr decompacted) during the Early Cretaceous is calculated on the basis of upper and lower age bounds. The termination of basin development is unknown, but deposition persisted into Miocene time in the nearby Nanyang, Niudingshan, and Lixian basins of the western Qinling Shan (Figure 2) [Zhai and Cai, 1984].
 For the lower Xining-Minhe basin, an average rate of 25 m/Myr (31 m/Myr decompacted) during the Early Cretaceous is calculated on the basis of six polarity chrons. Although intermediate levels of the basin lack paleomagnetic data, underlying and overlying magnetostratigraphic constraints define an average rate of 4.1 m/Myr (5.6 m/Myr decompacted) during Late Cretaceous-early Paleocene deposition. Paleomagnetic data for the upper Xining-Minhe basin incompletely sample polarity chrons of Paleocene-Oligocene strata but the magnetostratigraphic correlation, which is partially restricted by palynomorph assemblages, yields an average rate of 10 m/Myr (13 m/Myr decompacted) for late Paleocene-Eocene time.
 On the basis of magnetic polarity stratigraphy and supporting palynological age data, a plot of age versus stratigraphic thickness (Figure 6) reveals a three-stage history of Mesozoic to mid-Cenozoic sediment accumulation in the Xining-Minhe basin: (1) initial rapid accumulation (Early Cretaceous), (2) substantially reduced accumulation (Late Cretaceous-Paleocene), and (3) renewed rapid accumulation (Paleogene). The Dangchang basin recorded a single phase of rapid accumulation (Early Cretaceous). Calculation of maximum and minimum decompacted thicknesses for the Xining-Minhe basin (based on a variable 500- to 1000-m-thick Oligocene-Miocene overburden and uniform 200-m-thick Pliocene-Quaternary loess) generates an envelope of possible accumulation histories (Figure 6). Total subsidence and tectonic subsidence histories [e.g., Hendrix et al., 1992; Sobel, 1999; Liu et al., 2001; Yin et al., 2002] are not calculated due to a lack of independent evidence for surface elevation through time. This reasoning follows Métivier et al.  in that accumulation need not be attributed to subsidence, particularly in topographically closed basins. Stages 1–3 of the sediment accumulation history are discussed below.
5.1. Basin Development: Stages 1 and 2
 Initial rapid accumulation (Stage 1) followed by decreased accumulation (Stage 2) is attributed to fault-controlled subsidence due to extension and subsequent postrift thermal subsidence. A variety of simple thermal models [McKenzie, 1978; Salveson, 1978] predict an exponential decay in subsidence rates for several tens of millions of years following a short-lived episode of extension, broadly matching the Cretaceous-Paleocene accumulation pattern for the Xining-Minhe basin (Figure 6).
 Paleocurrent data, grain-size trends, and potential basin margin faults are consistent with initial fault-controlled subsidence during Early Cretaceous time. Paleocurrent measurements (Figure 7) of trough cross strata, imbricated clasts, and primary current lineations for the Xining-Minhe basin suggest principal sediment sources to the south and east, consistent with a general coarsening of facies toward the south and east. Two faults displaying normal separation have been mapped in this region (Figure 7) along the southern and eastern margins of the basin: a north dipping fault and west dipping fault, respectively [QBGMR, 1985; Gansu Bureau of Geology and Mineral Resources (GBGMR), 1988]. These faults may form a linked set of faults in a sinistral transtensional setting, similar to the hypothesis of Vincent and Allen  for Jurassic-Cretaceous basin development in the Hexi Corridor (Figure 2). Paleocurrents for the Dangchang basin (Figure 7) indicate northeast directed transport, consistent with a profound decrease in grain size from the southwestern limb to northeastern limb of the Dangchang syncline (Figure 5). This dispersal pattern is tentatively attributed to motion along a zone of mapped structures on the southwestern margin of the basin (Figure 7). Further work is needed to assess the original magnitude and sense of slip on this fault system.
 Basin development linked to extensional faulting and postrift thermal subsidence can be reconciled with recent thermochronological results for the northern Tibetan Plateau [George et al., 2001; Jolivet et al., 2001; Sobel et al., 2001; Delville et al., 2001; Cowgill et al., 2003]. Although the different data sets suggest slightly different time frames, they consistently indicate rapid Jurassic cooling (either a single cooling episode or series of episodes) and negligible cooling to even minor reheating during Cretaceous time. Minor reheating [e.g., Jolivet et al., 2001] may be the result of burial beneath 1–2 km of Cretaceous synrift and postrift strata.
 Increased early to mid-Cenozoic sediment accumulation (Stage 3) is revealed by magnetic polarity stratigraphy and palynology for the Xining-Minhe basin, and other regional considerations. An initial increase during Paleogene time was followed by accumulation of a 500- to 1000-m-thick Oligocene-Miocene succession [QBGMR, 1985; GBGMR, 1988]. Although not directly dated, the magnitude of Oligocene-Miocene deposition requires a further increase in accumulation rates over the Paleogene rates (Figure 6). An extensive Pliocene-Quaternary loess constitutes an additional ∼200 m of accumulation [Sun and Liu, 2000].
 Paleocurrent data (Figure 7) are consistent with multiple localized sediment source areas during Cenozoic time. Paleogene strata from the southern Longzhong basin indicate sediment transport from south to north. To the south, relatively small basins in the western Qinling Shan (Figure 7) contain evidence for both south and north directed transport, suggesting diverse local sediment sources. In addition, lateral facies changes and local correlations help define the initial expression of ranges that partitioned the larger Xining-Minhe basin into smaller entities such as the Xining, Lanzhou, Linxia, and Longzhong-Jingning subbasins of the Longzhong basin (Figure 2). Specifically, uplift of two west-northwest striking ranges can be evaluated: the Laji Shan (south of Xining) and the Maxian Shan (south of Lanzhou). The Cretaceous Hekou and Minhe groups can be confidently correlated across these ranges. In contrast, Eocene-Oligocene strata are well developed north of the Laji Shan in the Xining subbasin but absent south of the range, where deposition commenced during Miocene time in the Gonghe, Guide, and Xunhua basins [Pares et al., 2003]. Similarly, initial expression of the Maxian Shan during roughly mid-Tertiary time is evidenced by contrasting Oligocene-Miocene lithofacies systems to the north and south of this range. Progressive compartmentalization of these <30,000 km2 basins by compressional and transpressional range uplift may provide a conceptual analogue for Cenozoic partitioning of the >100,000 km2 basins of the Qaidam and Qilian Shan regions [Bally et al., 1986; Meyer et al., 1998].
 Published thermochronological data support this history of localized Tertiary rock uplift. Evidence for Eocene-Miocene cooling is present in piecemeal fashion from the western Tarim basin, Altyn Tagh region, and Qilian Shan [Sobel and Dumitru, 1997; Mock et al., 1999; George et al., 2001; Jolivet et al., 2001; Sobel et al., 2001; Cowgill et al., 2003]. Tertiary cooling ages occur in smaller proportions and in more restricted regions than the regionally extensive Jurassic cooling ages, consistent with local range uplift during Tertiary time. On the basis of the accumulation rates, thermochronological data, and possible age of seismically imaged growth strata in the Qaidam basin [Bally et al., 1986], these localized uplifts may have initiated as early as ∼40 Ma and are reasonably attributed to shortening induced by the Indo-Asian collision. We propose that range uplift in Paleogene compressional and transpressional settings generated sufficient topographic loads to generate new flexural basins and drive renewed subsidence in preexisting basins, although sediment ponding due to closure of drainage outlets may have accentuated this process [Métivier et al., 1998; Tapponnier et al., 2001].
 Distinguishing Mesozoic from Cenozoic strata remains a fundamental problem in the Tibetan plateau. Paleomagnetic studies combined with fossil age control help restrict possible stratigraphic ages. New magnetostratigraphic and palynological biostratigraphic data are reported for Upper Jurassic-Lower Cretaceous to mid-Tertiary successions in the Xining-Minhe and Dangchang basins of the northeastern Tibetan plateau. Although complete identification of magnetic polarity zones is not realized, the magnetostratigraphic correlations presented are partially restricted by palynological age constraints. Accumulation rates vary from relatively high values during Early Cretaceous time to low values during Late Cretaceous-Paleogene time, consistent with regional thermochronological and structural evidence for Late Jurassic-Early Cretaceous extension and Late Cretaceous-Paleogene postrift thermal subsidence. Renewed increases in accumulation rates in the Xining-Minhe basin during early to mid-Tertiary time require an additional, younger mechanism of basin subsidence. Published thermochronological evidence for exhumation of localized ranges and regional paleomagnetic evidence for a vertical axis clockwise rotation at 40–30 Ma [Dupont-Nivet et al., 2004] suggest that flexural subsidence related to individual range uplifts drove increased early to mid-Tertiary accumulation in the northeastern Tibetan Plateau during the initial Indo-Asian collision.
 Funding provided by American Chemical Society (Petroleum Research Fund) grant PRF-37207-G8 (Horton), National Science Foundation grant EAR-0106677 (Horton), and a Geological Society of America research grant (Dupont-Nivet). We thank Huihua Zhang and Xiaowei Jiang (Chinese Academy of Sciences, Institute of Geochemistry, Guangzhou) for logistical and field assistance. Comments by Randy Enkin, Bradley Ritts, and an anonymous reviewer improved the manuscript.