Objective. To address principles in the design and conduct of clinical trials on prostate cancer (PC) with special reference to localized disease. Methods. In advance of and during the World Health Organization (WHO) conference on Prostate Cancer in Stockholm in September 1996, 6 members of a working group evaluated and reached consensus on key points for the planning and conduction of controlled clinical trials in PC. The key points discussed were 1) hypothesis formulation, 2) general methodological principles, 3) special problems of PC trials, 4) alternatives to randomization, and 5) trial organization. Results. The hypothesis must be clearly formulated and also clinically relevant enough to justify the expenses (in a broad sense) of a randomized clinical trial. Patient selection, definition of endpoints, and sample size calculations must be carefully considered and correspond to the aims of the study. Stratification on important prognostic factors should be contemplated. Maintaining the accrual rate and ensuring compliance are critical for a quality study. Survival is the main endpoint and intention to treat analysis is the standard methodology. Secondary endpoints (eg, quality of life and costs) are important for the evaluation of many treatment modalities. The use of surrogate endpoints for survival, such as prostate- specific antigen (PSA) elevation, may be misleading. Surrogate endpoints require further validation. Special features, such as long natural history in localized disease and difficulties in assessing objective responses in advanced disease, need consideration in PC trials. The controlled randomized trial is the gold standard methodology. Nonrandomized trials are often hampered by severe methodological problems, such as selection bias and biased ascertainment of endpoints due to the doctor's or patient's preferences. For the organization of a successful trial, a well-constructed study protocol is essential. Good clinical practice as defined within the European Community helps define and solve practical problems. Conclusions. The past 30 years of poorly designed clinical research on PC has left us without reliable answers in key clinical issues regarding, for example, the efficacy of primary treatment of localized PC. It is a stimulating and important challenge to the urological scientific community to conduct well-organized controlled clinical trials.