The risk of intracranial tumors after exposure to ionizing radiation during infancy has been studied in a pooled analysis of two Swedish hemangioma cohorts (n = 28,008). The mean absorbed intracranial dose was low (7 cGy, range 0-11.5 Gy). The cohorts were followed up in the Swedish Cancer Register for incident intracranial tumors during the period 1958-1993. Eighty-eight tumors were found in 86 individuals compared to 60.72 expected [standardized incidence ratio (SIR) 1.42, 95% confidence interval (CI) 1.13-1.75]. The SIR increased significantly in ascending dose categories (P = 0.02). Dose-response analyses were performed with Poisson regression methods. There was a significant effect of dose, and the dose-effect relationship was negatively modified by age at first treatment. This indicates a higher risk for those exposed earlier in life. A linear dose-response model modified by age at first treatment resulted in the best fit. The excess relative risk (ERR) was 2.7/Gy (95% CI 1.0-5.6). The ERR/Gy was 4.5 if the treatment was given before 5 months of age, 1.5 if it was given at 5-7 months and 0.4 if it was given later. The study thus strongly indicates that there exists a dose-response relationship between absorbed dose in the brain and the subsequent risk of developing an intracranial tumor and that the risk is higher among infants exposed at younger ages.