To analyze the probability of recovery of spermatogenesis after orchidectomy and cisplatin-based chemotherapy (CT) for testicular germ cell cancer. One hundred seventy-eight patients treated between 1979 and 1991 were selected by the requirement of sperm count both pre-CT and post-CT. Counts were classified as normospermic (NS) if more than 10 x 10(6)/mL, oligospermic (OS) if 1 to 9 x 10(6)/mL, and azoospermic (AS) if less than 1 x 10(6)/mL. The median follow-up time after CT before sperm analysis was 30 months. Analysis of 170 patients whose spermatogenesis was reassessed at least 1 year after CT showed that of 89 patients whose pre-CT counts were NS, the post-CT count was NS in 64%, OS in 16%, and AS in 20%. There was clear evidence for continued recovery beyond 1 year; the probability of spermatogenesis increased to 48% by 2 years and 80% by 5 years. There was a significantly higher probability of recovery to OS and NS count levels in the 54 patients treated with carboplatin-rather than cisplatin-based therapy. There was an independent and similar effect of normal pre-CT count. There was a reduced probability to recover to OS in the 26 patients treated with more than four cycles of CT. A prognostic model identified three groups with 25%, 45%, and 82% probabilities of recovering spermatogenesis by 2 years after CT. Analysis of pre-CT sperm count together with details of planned treatment can be used to predict recovery of spermatogenesis following germ cell CT.