This study evaluated male gonadal function in long-term survivors of childhood cancer and assessed the suitability of offering sperm analysis to all those patients independently of the diagnosis and treatment received. A total of 43 survivors of acute lymphoblastic leukemia (21), acute myeloid leukemia (1), neuroblastoma (8), ganglioneuroblastoma (1), ganglioneuroma (2), Wilms' tumor (9), and mesoblastic nephroma (1) underwent sperm analysis at a mean age of 20.2 years, after a mean time off treatment of 13.6 years. Eight of the patients (19%) were azoospermic, 2 (5%) were severely oligo-asthenozoospermic, and only 16 (37%) were normozoospermic. A control group of healthy volunteers aged < or = 30 years included no azoospermic subjects, 7% severely oligo-asthenozoospermic, and 67% normozoospermic. Comparisons were also made with patients treated at our Human Reproductive Unit aged < or = 30 years (n = 373) whose percentages for the above parameters were 4, 9, and 42%, respectively. Cumulated cyclophosphamide dose and basal follicle-stimulating hormone (FSH) levels were identified as independent factors associated with azoospermia or severe oligo-asthenozoospermia. Azoospermic and severely oligo-asthenozoospermic survivors had significantly smaller mean testicular volume and higher basal FSH levels than the other survivors, but small testicles (sum of both testicular volume < or = 20 mL) and/or abnormally high basal FSH (> 10 mIU/mL) were present in only half of the azoospermic survivors. Male long-term survivors of childhood cancer constitute a high-risk subpopulation for altered sperm analysis. It seems justified to offer sperm analysis to all long-term survivors.