OBJECTIVE To compare the efficacy and safety of insulin aspart (IAsp) and human insulin (HI) when applied as meal-time insulin with neutral protamine Hagedorn insulin (NPH) at bedtime in diabetics. METHODS A total of 220 Chinese subjects with type 1 or type 2 diabetes from 5 different hospitals were randomized by a ratio of 1:1 into two groups accepting IAsp or HI combined with NPH respectively. The main endpoints were assessed by fasting plasma glucose (FPG), 2 hour postprandial plasma glucose (2 h PPG), HbAlc and hypoglycemia. RESULTS A greater reduction in mean 2 h PPG was achieved in the IAsp group [(14.6 +/- 5.3) mmol/L] as compared with the HI group [(8.4 +/- 4.1) mmol/L] (P < 0.01, adjusted for baseline value, center effect and diabetes type). Significantly more IAsp-treated subjects reached the 2 h PPG target (50.0% vs 25.5%, P < 0.01). HbA1c was reduced more in IAsp/NPH group [(9.3 +/- 1.4)% vs (7.7 +/- 1.3)%] than in HI/NPH group [(9.2 +/- 1.2)% vs (7.7 +/- 1.2)%]. HbA1c target was reached by 24.5% (IAsp) vs 14.5% (HI) of subjects (P < 0.05). No major hypoglycemia or serious adverse events were observed for the IAsp group. Lower incidence of nocturnal hypoglycemia (IAsp/NPH: 3% vs HI/NPH: 4%) was reported in the IAsp group. Average daily insulin doses were 0.60/0.23 (IAsp/NPH) and 0.65/0.24 (HI/NPH) IU/kg respectively. CONCLUSIONS Treatment of IAsp in basal-bolus therapy in combination with NPH provides a superior postprandial glucose control and allows more subjects to reach the glycemic target without elevating the nocturnal hypoglycemic risk or adverse events.