AIMS: The basal insulin analogue LY2605541, a PEGylated insulin lispro with prolonged duration of action, was previously shown to be associated with modest weight loss in Phase 2, randomized, open-label trials in type 2 (N=288) and type 1 (N=137) diabetes mellitus (T2DM and T1DM), compared with modest weight gain with insulin glargine. Exploratory analyses were conducted to further characterize these findings. METHODS: Pearson correlations between change in body weight and other variables were calculated. Continuous variables were analysed using a mixed linear model with repeated measurements. Proportions of subjects with weight loss were analysed using Fisher's exact test for T2DM and Nagelkerke's method for T1DM. RESULTS: Weight loss was more common in LY2605541-treated patients than in patients treated with insulin glargine (T2DM: 56.9 vs. 40.2%, p=0.011; T1DM: 66.1 vs. 40.3%, p<0.001). More LY2605541-treated patients experienced >/=5% weight loss compared to patients treated with glargine (T2DM: 4.8 vs. 0%, p=0.033; T1DM: 11.9 vs. 0.8%, p<0.001). In both the T1DM and T2DM studies, weight change did not correlate with baseline body mass index (BMI), or change in HDL-cholesterol in either treatment group. No consistent correlations were found across both studies between weight change and any of the variables assessed; however, weight change was significantly correlated with hypoglycaemia rate in glargine-treated T2DM patients. CONCLUSION: In two Phase 2 trials, improved glycaemic control with long-acting basal insulin analogue LY2605541 is associated with weight loss in previously insulin-treated patients. This weight change is independent of baseline BMI or hypoglycaemia.