The paper discusses the use of Global Positioning System (GPS) receivers for remote sensing of atmospheric water vapor. The authors present a method to transform observed zenith wet delays (ZWD) into estimates of precipitable water (PW) by multiplying the ZWD by a constant factor, which depends on the refractivity of the atmosphere and the weighted mean temperature. The mean temperature must be estimated a priori, and the paper shows that the relative error in this transformation closely approximates the relative error in the predicted mean temperature, which can be estimated using numerical weather models with an rms relative error of less than 1%. The authors also analyze the uncertainties in the refractivity constants and conclude that the dominant source of error in transforming ZWD into PW estimates is the accuracy of the a priori mean temperature estimate. They find that combining their previous estimate of ZWD errors with the error introduced during the transformation results in an overall error in PW estimates of less than 2 mm + 1% of the PW, with long-term biases of less than 2 mm. The paper emphasizes the importance of improving techniques for estimating ZWDs and suggests that better determination of refractivity constants could contribute to more accurate PW estimates.The paper discusses the use of Global Positioning System (GPS) receivers for remote sensing of atmospheric water vapor. The authors present a method to transform observed zenith wet delays (ZWD) into estimates of precipitable water (PW) by multiplying the ZWD by a constant factor, which depends on the refractivity of the atmosphere and the weighted mean temperature. The mean temperature must be estimated a priori, and the paper shows that the relative error in this transformation closely approximates the relative error in the predicted mean temperature, which can be estimated using numerical weather models with an rms relative error of less than 1%. The authors also analyze the uncertainties in the refractivity constants and conclude that the dominant source of error in transforming ZWD into PW estimates is the accuracy of the a priori mean temperature estimate. They find that combining their previous estimate of ZWD errors with the error introduced during the transformation results in an overall error in PW estimates of less than 2 mm + 1% of the PW, with long-term biases of less than 2 mm. The paper emphasizes the importance of improving techniques for estimating ZWDs and suggests that better determination of refractivity constants could contribute to more accurate PW estimates.