In this study, we prepared several shortened and full-length insulin analogues with substitutions at position B26. We compared the binding affinities of the analogues for rat adipose membranes with their ability to lower the plasma glucose level in nondiabetic Wistar rats in vivo after subcutaneous administration, and also with their ability to stimulate lipogenesis in vitro. We found that [NMeHisB26]-DTI-NH 2 and [NMeAlaB26]-DTI-NH 2 were very potent insulin analogues with respect to their binding affinities (214 and 465%, respectively, compared to that of human insulin), but they were significantly less potent than human insulin in vivo. Their full-length counterparts, [NMeHisB26]-insulin and [NMeAlaB26]-insulin, were less effective than human insulin with respect to binding affinity (10 and 21%, respectively) and in vivo activity, while [HisB26]-insulin exhibited properties similar to those of human insulin in all of the tests we carried out. The ability of selected analogues to stimulate lipogenesis in adipocytes was correlated with their biological potency in vivo. Taken together, our data suggest that the B26 residue and residues B26-B30 have ambiguous roles in binding affinity and in vivo activity. We hypothesize that our shortened analogues, [NMeHisB26]-DTI-NH 2 and [NMeAlaB26]-DTI-NH 2, have different modes of interaction with the insulin receptor compared with natural insulin and that these different modes of interaction result in a less effective metabolic response of the insulin receptor, despite the high binding potency of these analogues.