Long baroclinic Rossby waves are potentially important in the adjustment of the tropical Pacific pycnocline to both annual and interannual wind stress curl fluctuations. Evidence for such waves is found in variations of the depth of the 20°C isotherm in the northern tropical Pacific during 1970–1987. A total of 199,067 bathythermograph profiles have been compiled from the archives of several countries; the data coverage is dense enough that westward propagating events may be observed with a minimum of zonal interpolation. After extensive quality control, 20°C depths were gridded with a resolution of 2° latitude, 5° longitude, and bimonths; statistical parameters of the data were estimated. A simple model of low-frequency quasi-geostrophic pycnocline variability allows the physical processes of Ekman pumping, the radiation of long (nondispersive) Rossby waves due to such pumping in midbasin, and the radiation of long Rossby waves from the observed eastern boundary pycnocline depth fluctuations. Although the wind stress curl has very little zonal variability at the annual period in the northern tropical Pacific, an annual fluctuation of 20°C depth propagates westward as a long Rossby wave near 5°N and 14°–18°N in agreement with the model hindcast. Near the thermocline ridge at 10°N, however, the annual cycle across the basin is dominated by local Ekman pumping. The wave-dominated variability at 5°N weakens the annual cycle of geostrophic transport of the North Equatorial Countercurrenl in the western Pacific. El Niño events are associated with westerly wind anomalies concentrated in the central equatorial Pacific; up welling wind stress curl is generated in the extraequatorial tropics by these westerlies. Long up welling Rossby waves forced by this curl pattern were observed to raise the western Pacific thermocline well outside the equatorial waveguide in the later stages of El Niños, consistent with the simple long-wave model. The observations suggest that although simple reflection of the long Rossby waves from the western boundary is not the major process affecting subsequent development on the equator, it is likely that the extraequatorial waves play some role in setting the timescale of ENSO events.