Data on the incidence of insulin-dependent diabetes mellitus (IDDM) are presented. The study concerned the Belgrade population, aged 0-14, over the period from 1982 to 1992. A retrospective technique was used to record newly diagnosed cases of IDDM. Hospital records were used as the primary source, and the Population-based register as the secondary source of data. In total, 259 newly diagnosed cases were detected (the completeness of ascertainment was 90.5%). In the Population-based register only 165 cases were included, i.e. 63.7% of all detected IDDM patients, or 57.7% of all expected IDDM incidences cosses. It must be, therefore, concluded that the Register for diabetes mellitus has not been a very reliable source of data.