The objective of this study was to analyze the value of the nadir level of prostate-specific antigen (PSA) to predict androgen-independent progression (AIP) in metastatic prostate cancer patients after androgen deprivation therapy. In a group of 185 metastatic prostate cancer patients who received androgen deprivation therapy serum PSA was determined every three months until AIP occurred. Multiple regression analysis was performed to define independent clinical and PSA-related predictors of AIP. AIP was assumed to be present after two consecutive increases in serum PSA after the PSA nadir. Independent predictors of the duration of AIP-free survival (less than 12 months versus more than 12 months) were the extent of bone involvement (six or fewer hot spots versus more than six) with an odds ratio (O.R.) of 3.95, Gleason score (7 or less versus more than 7) with an O.R. of 3.47, and PSA nadir (2 microg/L or less versus more than 2 microg/L) with an O.R. of 14.63. AIP was independently predicted by the extent of bone involvement with an O.R. of 1.72, Gleason score with an O.R. of 1.74, PSA nadir with an O.R. of 3.22, and time to reach the PSA nadir (9 months or less versus more than 9 months) with an O.R. of 2.84. When patients were stratified according to these predictors, those with three good prognostic factors had a median AIP-free survival of 58 months while those with two, one or no good prognostic factors had a median AIP-free survival of 19 months, 12 months and 7 months, respectively. We conclude that the PSA nadir seems to be a good predictor of AIP in patients with metastatic prostate cancer after androgen deprivation therapy. Time to PSA nadir, extent of bone involvement and Gleason score are also independent predictors. The combination of these prognostic factors allows to stratify metastatic prostate cancer patients for the prediction of AIP.