In 224 consecutive patients with hormone-resistant prostatic cancer referred to 2 European Cancer Centres for palliation of painful bone metastases the one year survival for all patients was 24% (2-year survival: 7%). The median survival was 8 months. In univariate analyses the following prognostic factors were identified: performance status, serum creatinine, alkaline phosphatase, duration of response to primary hormone treatment, degree of bone scan involvement and hemoglobin. Multivariate analyses confirmed the four first parameters to be independent factors. A prognostic model was established (no or one risk factors vs 2 risk factors vs 3 or 4 risk factors) based on performance status, creatinine, alkaline phosphatase and hormone response duration. The median survival of these groups was 10 months, 6 months and 3 months, respectively. This model proved to be discriminative in an external data set of 214 patients with hormone-resistant prostatic cancer entered in two prospective trials. The above differences in outcome between readily and simply defined prognostic groups are greater than the differences one can realistically hope to produce using new treatment strategies. These prognostic factors should be taken into account both in the design and interpretation of clinical studies dealing with the treatment of hormone-resistant progressing prostatic cancer and painful bone metastases.