Clinical and laboratory data of 1248 newly diagnosed diabetic children at the time of diagnosis were analysed. All children were admitted to the University Children's Hospital in Sofia, 45.9% before first their insulin injection. Symptoms preceding the diagnosis and laboratory data (plasma glucose and ketonuria) were analysed, respectively for 1100 and 1022 children. Blood pH (mainly arterialized) was available in 558 ketonuria positive children and in other 82 acidotic breathing was reported. Mother's education was noted in the 1226 hospital records. Among the children with known urinanalysis 13.5% were without ketonuria (148 patients), 12 of them with fasting blood glucose < or = 6.4 mmol/1 (115 mg/dl). Eighteen-point-two percent of all children were hospitalized in a state of severe ketoacidosis (blood pH < or = 7.2 or reported acidotic breathing). The average duration of thirst and polyuria was 28 +/- 33 days. Ketonuria negative children with plasma glucose < or = 6.4 mmol/l showed a significantly shorter period of symptoms, compared to those with plasma glucose > 6.4 mmol/1 (17 +/- 25 vs. 25 +/- 31 days; P = 0.0991). The cases with severe ketoacidosis, compared to those with mild ketoacidosis (blood pH 7.21 - 7.34) showed shorter period of symptoms too (P = 0.0658). Moderate positive relation existed between the age at diagnosis and duration of symptoms (chi 2 = 43.28, D.F. = 8, P = 0.0000). The percentage of severe ketoacidosis is higher in the younger age groups. Febrile illness, preceding the start of the symptoms was more common in the groups with shorter duration of symptoms (up to 1 month), but did not change the proportion of severe ketoacidosis. No significant difference was found between the level of mother's education and duration of the symptoms before diagnosis (P = 0.9782). We conclude that the level of metabolic disturbances at the diagnosis of IDDM among children was not influenced by the duration of the preceding symptoms. The severity of clinical picture was possibly dependent on the degree of insulinopenia, i.e. the rate of beta-cell destruction. Clinical heterogeneity was possibly dependent on genetical heterogeneity related to HLA class II genes.