The European Center for Medium Range Weather Forecasts (ECMWF) level IIIb dataset is used to construct global pressure analyses of the tropopause surface during January 1979. Two methods are employed: a dynamical method based on isentropic potential vorticity (IPV) and a thermal method based on lapse rate criteria. Regional tropopause pressure analyses are extracted from the global analyses and compared against distributions derived from rawinsonde data. The coarse vertical resolution of the ECMWF data compromises the ability to resolve abrupt stability changes between the troposphere and stratosphere and impacts tropopause analyses using both methods. Sensitivity of the derived tropopause pressures to a range of IPV and lapse rate thresholds is examined. For the assimilated dataset employed herein, 3.5 IPV units represent an optimal value for tropopause analysis outside the tropics. Modification of the WMO lapse rate criteria does not significantly improve tropopause analysis globally. Both methods capture the large-scale features of the radiosonde-reported tropopause surface in the regional analyses, although each approach has limitations. The spatial structure and temporal evolution of the dynamically determined tropopause surface within a developing extratropical cyclone is found to be superior to that based on lapse rate criteria, while only the lapse rate method is a viable approach in the tropics. We conclude that the pressure of the tropopause surface can be determined globally using ECMWF assimilated data. The preliminary results are encouraging and suggest that it is feasible to proceed beyond sounding analyses and case studies for determining the tropopause position. We view this to be an important first step toward implementing global studies of stratospheric–tropospheric exchange.