Apparent changes in the local occurrence and regional extent of southwestern U.S. piñon–juniper woodlands since Euro-American settlement (i.e. historic expansion) are widely reported. These changes are commonly attributed to rapid onset of intensive and unregulated livestock grazing in western rangelands beginning ca 1850. However, other potential drivers of historic expansion including climatic warming, biological inertia, elevated CO2, and post-disturbance successional recovery have also been noted. Landscape patterns of woodland expansion have not been well characterized, thus the magnitude, extent, and timing of historic change remains uncertain and the relative importance of ecological drivers difficult to assess. Previous work within a monsoonal north-central New Mexico study area demonstrated the utility of logistic modeling to address these questions. Here I expand these efforts by sampling and modeling piñon-juniper woodlands across a climatically variable four-state area (i.e. New Mexico, Arizona, Colorado, and Utah). Using MARS, a piece-wise linear procedure, I model occurrence of pre- versus post-settlement aged stands in relation to environmental factors, predictively map woodland expansion patterns, and evaluate grazing versus other potential drivers of historic change. Pre-settlement woodlands were most commonly found in upland settings consistent with previous work, although this relationship was less pronounced in winter moisture areas on the Colorado Plateau. Post-settlement stands in contrast were typically associated with more productive valley-terrace and toe-slope settings. However, predicted extent of expansion regionally was largely restricted to summer moisture areas within the range of one-seed juniper. This dichotomy, where historic expansion was directional into depositional settings atypical of pre-settlement woodland occurrence supports the view of intensive grazing as a disruptive disturbance. However, the more limited expansion predicted for winter moisture areas suggests grazing effects were likely contingent on bio-climatic context including ecological potential of different juniper species and affected sites.