Wet tropospheric path delay can be a major error source for Global Positioning System (GPS) geodetic experiments. We investigate strategies for minimizing this error using data from CASA Uno, the first major GPS experiment in Central and South America, where wet path delays may be both high and variable. We compared wet path delay calibration using water vapor radiometers (WVRs) and residual delay estimation, with strategies where the entire wet path delay is estimated stochastically without prior calibration, using data from a 270 km test baseline in Costa Rica. Both approaches yield centimeter-level baseline repeatability and similar tropospheric estimates, suggesting that WVR calibration is not critical for obtaining high precision results with GPS in the CASA region.