The clinical course of prostate cancer (PCa), the most common cancer in Swedish men, is highly variable and difficult to predict. Consequently, there is an urgent need to distinguish tumours with a high risk of progression from those with a low risk. To investigate the prognostic implications of proliferation and apoptosis, two important processes in tumor biology, immunoreactivity for biomarkers associated with these processes was assessed, quantified in indexes, and related to cause-specific survival (CSS). A consecutive series of 186 men presenting with voiding symptoms and PCa were treated with transurethral resection and deferred endocrine therapy. After 13-21 years follow-up, 43% of these men had died of prostate cancer. In a subgroup of men with localised disease at the time of diagnosis, 27% succumbed to the disease. Immunoreactivity for p53 protein, indicative of a defective p53 function, predicted shorter CSS in univariate (52 vs 123 months, p < 0.0001), but not in multivariate analysis. Mean index for the apoptosis blocking bcl-2 protein was higher in foci of prostatic intraepithelial neoplasia, a putative precursor to PCa, than in manifest cancer areas (79 vs 12, p < 0.0001). This indicates that bcl-2 may be involved in early tumourigenesis. No prognostic value was found for the bcl-2 index. A high index for the proliferation marker Ki-67 predicted shorter CSS in univariate (53 vs 132 months p < 0.0001) and in multivariate analysis. To test if p53 is predictive for clinical radioresistance, as suggested by experimental models, p53 immunoreactivity was investigated in biopsies obtained before radical radiotherapy in an unrelated series of 60 PCa patients. Patients with p53 reactive tumours had longer CSS, indicating that p53 is not treatment-predictive for radiotherapy in Pca. Core biopsies were obtained before and a week after castration therapy in patients with advanced PCa. According to the serum prostate specific antigen (PSA) level 3 months after therapy, 15 responding tumours and 13 non-responding tumours were selected. Regressive morphology was seen in 14/15 responders after castration therapy, compared with 4/13 non-responders. Median apoptotic index increased significantly after castration therapy for responders (from 2.6 to 3.5, p < 0.05) whereas it was 2.8 before and after therapy in non-responders. This indicates that subsequent clinical response can be predicted by the induction of regressive morphology and an increase in apoptotic index. In conclusion, immunoreactivity for Ki-67 appeared to be a putative prognostic factor in PCa, whereas the prognostic value of p53 and bcl-2 was dubious. p53 immunoreactivity did not appear to be predictive of radioresistance in PCa. Cellular response in biopsies shortly after castration therapy might be treatment-predictive.