OBJECTIVES: This study aims to investigate the metabolic effects of biphasic insulin lispro 50/50 in routine clinical practice. A total of 229 patients who were ≥18 years old with diabetes, newly treated with biphasic insulin lispro 50/50, were sourced from six secondary care services in England. METHODS: Detailed clinical parameters were compared at baseline, and 3 and 6 months post-initiation. Responders was defined as those with HbA1c <7.5% (58 mmol/mol) and/or an HbA1c reduction of >1% (11 mmol/mol) at 6 months. RESULTS: HbA1c showed significant reduction: -0.93% (-10 mmol/mol) and -1.2% (-13 mmol/mol) at 3 and 6 months respectively, while no significant change was noted for all the other parameters. When analyzed according to frequencies of injections/day, the greatest reduction was observed with the three times a day regimen (-1.0% [-11.0 mmol/mol] and -1.3% [-14.6 mmol/mol] at 3 and 6 months respectively). HbA1c reduction was greatest in the group who previously received a basal-bolus insulin regimen: (-0.8% [-9.0 mmol/mol] and -1.5% [-16.2 mmol/mol] at 3 and 6 months respectively). Reduction in weight was observed at 3 months (-1.8 kg ± 4.3) only for those who were previously on a basal-bolus insulin regimen. Insulin doses increased following conversion to biphasic insulin lispro 50/50, irrespective of the types of insulin used prior to biphasic insulin lispro 50/50, but this was not associated with weight gain. The independent predictors of response to biphasic insulin lispro 50/50 were baseline HbA1c, Caucasian, presence of nephropathy, prior use of basal-bolus insulin and prior use of other premixed combination. CONCLUSION: Biphasic insulin lispro 50/50 is therefore an effective therapeutic option for achieving glycemic control in patients with suboptimal HbA1c levels, especially among those who were previously on a basal-bolus insulin regimen and those who received it three times daily, with a neutral effect on weight parameters. LIMITATIONS: This was a retrospective study of routine clinical practice and is therefore limited by allocation bias and some missing data. Information on rates of hypoglycemia and quality of life are not available.