The thyroid dose due to 131I releases during the Chernobyl accident was reconstructed for children and adolescents in two cities and 2122 settlements in Belarus, and in one city and 607 settlements in the Bryansk district of the Russian Federation. In this area, which covers the two high contamination spots in the two countries following the accident, data on thyroid cancer incidence during the period 1991-1995 were analysed in the light of possible increased thyroid surveillance. Two methods of risk analysis were applied: Poisson regression with results for the single settlements and Monte Carlo (MC) calculations for results in larger areas or sub-populations. Best estimates of both methods agreed well. Poisson regression estimates of 95% confidence intervals (CIs) were considerably smaller than the MC results, which allow for extra-Poisson uncertainties due to reconstructed doses and the background thyroid cancer incidence. The excess absolute risk per unit thyroid dose (EARPD) for the birth cohort 1971-1985 by the MC analysis was 2.1 (95% CI 1.0-4.5) cases per 10(4) person-year Gy. The point estimate is lower by a factor of two than that observed in a pooled study of thyroid cancer risk after external exposures. The excess relative risk per unit thyroid dose was 23 (95% CI 8.6-82) Gy(-1). No significant differences between countries or cities and rural areas were found. In the lowest dose group of the settlements with an average thyroid dose of 0.05 Gy the risk was statistically significantly elevated. Dependencies of risks on age-at-exposure and on gender are consistent with findings after external exposures.