3-Guanidinopropionic acid (1) has been demonstrated both to improve insulin sensitivity and to promote weight loss selectively from adipose tissue in animal models of non-insulin-dependent diabetes mellitus (NIDDM). However, 1 has also been shown to be a substrate for both the creatine transporter and creatine kinase, leading to marked accumulation in muscle tissue as the corresponding N-phosphate. The corresponding aminoguanidine analogue 2 was recently discovered to retain the antidiabetic activity of 1 while being markedly less susceptible to creatine-like metabolism, suggesting that it should have less potential to accumulate in muscle. Further structural modification of 2 was undertaken to investigate whether the antidiabetic potency could be augmented while maintaining resistance to creatine-like metabolism. Modifications such as alpha-alkylation, homologation, and bioisosteric replacement of the aminoguanidine all were detrimental to antidiabetic activity. However, the simple regioisomeric aminoguanidinoacetic acid 9 and diaminoguanidinoacetic acid analogue 7 were found to be equipotent to 2, leading eventually to the discovery of the significantly more potent diaminoguanidinoacetic acid regioisomers 52 and 53. Further attempts to modify the more active template represented by 52 led only to reductions in antidiabetic activity. Each of the new active analogues displayed the same resistance to creatine-like metabolism as 2. Further testing of 7, 9, and 53 in obese diabetic ob/ob mice confirmed that weight loss is induced selectively from adipose tissue, similar to the lead 1. Administration of 53 to insulin-resistant rhesus monkeys led to reductions in both fasting and post-prandial plasma glucose levels with concomitant reductions in plasma insulin levels, suggesting that the compound improved the action of endogenous insulin. Compounds 7 and 53 were selected for further preclinical development.