The objective of this 6-month, open-label, randomized, two-period crossover study was to compare glycemic control when patients were treated with (1) 2 manufactured premixed insulin formulations containing insulin lispro and a novel insulin lispro-protamine formulation, neutral protamine lispro (NPL), and (2) 2 manufactured premixed human insulin formulations, human insulin 50/50 and human insulin 30/70. One hundred individuals, 37 with type 1 diabetes mellitus (12 females, 25 males; mean age, 39.4 years; mean body mass index [BMI], 24.8; mean duration of diabetes, 12.9 years) and 63 with type 2 diabetes mellitus (33 females, 30 males; mean age, 59.0 years; mean BMI, 28.4; mean duration of diabetes, 12.6 years), were treated with insulin lispro mixtures. Insulin lispro Mix50 (50% insulin lispro/50% NPL) and human insulin 50/50 (50% regular insulin/50% neutral protamine Hagedorn [NPH] insulin) were administered before breakfast; insulin lispro Mix25 (25% insulin lispro/75% NPL) and human insulin 30/70 (30% regular insulin/70% NPH) were administered before dinner. Blood glucose (BG), hypoglycemic episodes (hypoglycemic signs or symptoms or BG <3.0 mmol/L), insulin dose and timing of dose before meals, and hemoglobin A1c were measured. Mean doses of insulin lispro and human insulin mixtures were similar overall and for both diabetes subgroups. However, compared with human insulin mixtures, twice-daily administration of insulin lispro mixtures resulted in improved postprandial glycemic control, similar overall glycemic control, and less nocturnal hypoglycemia, as well as offering the convenience of dosing closer to meals.