Aims: To investigate pain, leakage and bleeding when injecting insulin with different diameters of needles. Methods: Sixty children and adolescents aged 9–21 yrs participated in study A and 40 aged 8–20 yrs in study B. Both were double‐blind and randomized. In study A, we evaluated the pain when injecting with three needles [Novo 27G/13 mm (N27), B‐D MicroFine IV 28G/13 mm (B28), NovoFine 28G/12 mm (N28)] and in study B, with three needles [NovoFine 28G/12 mm (N28), B‐D MicroFine+ 29G/13 mm (B29), NovoFine 30G/8 mm (N30)] and one placebo injection (no needle mounted). Abdominal and thigh injections were given in a 45° angle with a lifted two‐finger skinfold on two different visits, scoring pain on a 10‐cm visual analog scale (VAS), and in study B faces were added to the scale. Results: The median VAS scores in study A were 1.2 cm (N27), 1.2 cm (B28) and 1.0 cm (N28) for abdominal injections, and 1.2 cm (N27), 0.7 cm (B28) and 1.1 cm (N28) (n.s.) for thigh injections. The median VAS scores in study B were 2.5 cm (N28), 2.3 cm (B29) and 2.8 cm (N30) (n.s.) for abdominal injections, and 2.0 cm (N28), 1.5 cm (B29) and 1.9 cm (N30) (n.s.) for thigh injections. The overall median score of placebo injections was 0.1 cm (p=0.0001). Bleedings were less common with the B29 needle (35.5%) than with the N28 needle (48.1%) (p=0.028) but with no difference compared to the N30 needle (39.2%). Leakage of insulin was found in 14% of abdominal and 25% of thigh injections (p=0.0001) with no difference between the needles. VAS scores were higher in study B which may be explained by the facial VAS scale increasing the range of answers. Conclusions: We found no difference in injection pain, preference, bleeding or insulin leakage between the needles. Decreasing the needle diameter from 0.4 to 0.3 mm (27–30G) does not seem to decrease pain perception in this age group.