We report a Global Positioning System (GPS) baseline determination which includes bias fixing (carrier phase ambiguity solutions) and tropospheric corrections based on simultaneous water vapor radiometer (WVR) observations. Measurements were obtained using single-wavelength codeless GPS receivers and dual-wavelength WVRs at each end of a 22-km baseline near Boulder, Colorado, during three evenings in July 1983. Local thunderstorm activity prevailed during these observations. The WVRs were pointed toward each satellite sequentially at 15-min intervals during GPS data aquisition. Radio path delay corrections as large as 10 cm were estimated from the WVR observations. Baseline vectors were determined using tropospheric corrections based on WVR and surface meteorological (met) measurements and by resolving phase ambiguities (bias fixing). A repeat measurement of the same baseline was performed without WVRs during clear weather conditions in September 1984. If WVR corrections were not applied, we found that time varying corrections based on surface met measurements improved baseline length repeatability by a factor of 2, compared to corrections based on default surface met values. Applying WVR corrections in bias-fixed baseline solutions, we found a 4-cm decrease in relative height and a corresponding factor of 3 improvement in the repeatability (to 0.6 ppm). The WVR corrections also enhanced our ability to resolve phase ambiguities. Bias-free solutions, with the WVR correction, showed essentially the same 4-cm decrease in the vertical component (with the repeatability improved to 1 ppm). Baseline solutions using simulated data that include only the WVR correction confirm this decrease in the vertical. These results suggest that during disturbed weather conditions WVR (and surface P and T) measurements may be useful in achieving reliable bias fixing as well as optimum vertical accuracy.