BACKGROUND Survival in patients with metastatic, chemotherapy-naive, androgen-independent prostate cancer (AIPC) is improved with 10 to 12 cycles of docetaxel-containing chemotherapy but further management is undefined. In the current study, the authors examined retreatment with the same regimen after a treatment holiday. METHODS Patients treated with docetaxel at a dose of 36 mg/m(2) plus either high-dose calcitriol (DN-101; 45 mug) or placebo administered weekly for 3 of every 4 weeks could suspend treatment if their serum prostate-specific antigen (PSA) level was reduced >or=50% and reached a level <or=4 ng/mL. PSA was monitored every 4 weeks (computed tomography scans were administered every 8 weeks in patients with measurable disease) during the treatment holiday. Treatment was resumed when the serum PSA rose by >or=50% and was >or=2 ng/mL or when there was other evidence of disease progression. The study was not powered to compare treatment holiday outcomes between the 2 arms. RESULTS A total of 250 patients were randomized 1:1. Overall, 18% of patients (20% in the high-dose calcitriol group and 16% in the placebo group) entered the intermittent chemotherapy arm. The median duration of the first chemotherapy holiday was 18 weeks (range, 4%70 weeks). On resumption of treatment after the first holiday, 45.5% of evaluable patients responded with a >or=50% reduction in serum PSA from their postholiday baseline, 45.5% met the criteria for stable PSA for at least 12 weeks, and 9.1% of patients developed disease progression. CONCLUSIONS To the authors' knowledge, the current study is the first report of intermittent chemotherapy in patients with AIPC who were prospectively tested in a large multi-institutional trial. This strategy results in a clinically significant duration of chemotherapy holidays and can be offered to a minority of patients. At the time of retreatment, the majority of patients again respond to treatment or their PSA levels stabilized. Additional studies of intermittent chemotherapy are needed to better characterize the optimal patient population and the optimal approach.