Insulin regimens and metabolic control in children and adolescents with Type 1 diabetes mellitus were evaluated in a cross-sectional, non-population-based investigation, involving 22 paediatric departments, from 18 countries in Europe, Japan, and North America. Blood samples and information were collected from 2873 children from March to August 1995. HbA 1c was determined once and analysed centrally (normal range 4.4-6.3 %, mean 5.4 %). Year of birth, sex, duration of diabetes, height, body weight, number of daily insulin injections, types and doses of insulin were recorded. Average HbA 1c in children under 11 years was 8.3 ± 1.3 % (mean ± SD) compared with 8.9 ± 1.8 % in those aged 12-18 years. The average insulin dose per kg body weight was almost constant (0.65 U kg -1 24 h -1 ) in children aged 2-9 years for both sexes, but there was a sharp increase during the pubertal years, particularly in girls. The increase in BMI of children with diabetes was much faster during adolescence compared to healthy children, especially in females. Sixty per cent of the children (n = 1707) used two daily insulin injections while 37 % (n = 1071) used three or more. Of those on two or three injections daily, 37 % used pre-mixed insulins, either alone or in combination with short-and intermediateacting insulin. Pre-adolescent children on pre-mixed insulin showed similar HbA 1c levels to those on a combination of short-and long-acting insulins, whereas in adolescents significantly better HbA 1c values were achieved with individual combinations. Very young children were treated with a higher proportion of long-acting insulin. Among adolescent boys, lower HbA 1c was related to use of more short-acting insulin. This association was not found in girls. We conclude that numerous insulin injection regimens are currently used in paediatric diabetes centres around the world, with an increasing tendency towards intensive diabetes management, particularly in older adolescents. Nevertheless, the goal of near normoglycaemia is achieved in only a few.