This study aimed to evaluate a simplified minimal model protocol for measuring insulin sensitivity in mild and severe type 2 diabetes, considering that changes in serum insulin during an insulin-modified intravenous glucose tolerance test almost only reflect the insulin injection. Two groups of diabetics treated with high doses of antidiabetic agents were recruited. Mean insulin responses were calculated in group 1 (n = 30). In group 2 (n = 38), we compared insulin sensitivity (SI) obtained with reference protocol with SI calculated by a minimal model procedure including the theoretical average insulin profile determined in group 1, and with Homeostasis Model Assessment (HOMA-R). Additionally, the cost of each procedure was calculated. SI measured by the reference method strongly correlated with SI determined by the simplified protocol (r = 0.966, p < 0.0001), while no correlation was found with HOMA-R (r = - 0.349, NS). Reduction of cost for HOMA-R and simplified minimal model procedure were - 92 and - 81 %, respectively. This simplified and relative inexpensive protocol, using minimal model procedure without insulin measurement, accurately measures SI regardless of beta-cell defect degree. This approach could be of interest when limits of validity of simple indexes are reached.