The increase in the incidence of prostate cancer observed over the past 2 decades is suggested to be largely due to increased detection of insignificant tumors. To explore this assumption, we investigated trends in survival of prostate cancer patients in southeastern Netherlands, an area with almost 1 million inhabitants, where the age-adjusted incidence of prostate cancer increased by 53% between 1971 and 1989, i.e., before the introduction of prostate-specific antigen testing. Survival was calculated for all patients registered in the Eindhoven Cancer Registry between 1971 and 1989 (n = 2,562). In spite of earlier diagnosis, survival barely changed during this time period. Five-year relative survival improved slightly from 53% [95% confidence interval (CI) 47%, 59%] in 1975-1979 to 56% (CI 51%, 61%) in 1985-1989. Stratified analyses suggested an improvement since 1980 for patients below 75 years with localized tumors but, despite possible stage migration, decreased survival for those with metastasized and/or poorly differentiated tumors. Patients below 75 years whose tumors were diagnosed unexpectedly during transurethral resection of the prostate (TURP) exhibited a relative survival of 85% 5 years and 68% 10 years after diagnosis. Less extensive application of TURP in The Netherlands might explain why our findings do not agree with those in Sweden and the United States. Inference from country-specific trends in survival appears not to be necessarily generalizable to other countries with a similar increase in the incidence of prostate cancer. We conclude from our results that earlier diagnosis of prostate cancer between 1971 and 1989 may be accompanied by an increased incidence of an aggressive variant.