Aims/hypothesis. The aim of this study was to screen part of the putative promoter sequence in addition to 14 potential phosphotyrosine residues of human IRS-2 for genetic variability which might cause changes in protein expression or function. Furthermore, the potential impact on insulin secretion and sensitivity of a previously identified IRS-2 variant (Gly1057Asp) was analysed Methods. The screenings were carried out by the SSCP-heteroduplex technique on DNA from Type II (non-insulin-dependent) diabetic patients. The impact of the Gly1057Asp variant was analysed in four glucose-tolerant Scandinavian study groups. Results. The results showed no nucleotide substitutions in the promoter sequence, however, a novel heterozygous amino acid variant was identified (Leu647Val). In an association study, the new variant was found in 3 of 413 diabetic patients and in none of 280 glucose tolerant subjects. The variant did not affect the binding of IRS-2 to the insulin receptor or p85α of phosphatidylinositol 3-kinase when measured in the yeast two-hybrid system. Examination of the common Gly1057Asp variant in 363 young healthy subjects and in 228 glucose tolerant offspring of one diabetic parent showed no differences in insulin secretion or insulin sensivity after an intravenous glucose tolerance test. Glucose tolerant middle-aged subjects homozygous for the polymorphism (n = 31), however, had on average a 25 % decrease in fasting serum insulin concentrations (p = 0.009) and 28 % (p = 0.01) and 34 % (p = 0.003) reductions in serum insulin concentrations at 30 and 60 min, respectively, during an OGTT compared with wildtype carriers (n = 107). In a cohort of 639 elderly Swedish men the amino acid variant did not have any detectable impact on insulin secretion after an OGTT. Conclusion/interpretation. No genetic variability was found in the IRS-2 promoter. A rare IRS-2 variant at codon 647 has been identified in Type II diabetic patients. The prevalent codon 1057 polymorphism had no consistent effect on insulin secretion or insulin sensitivity. [Diabetologia (1999) 42: 1244–1249]