AIMS:To investigate the association between polymorphisms of the aldose reductase gene and diabetic nephropathy in both Type 1 and Type 2 diabetes mellitus, and to carry out a meta-analysis of published results. METHODS:We have investigated the role of two aldose reductase polymorphisms in four independent cohorts of cases and controls (two each with Type 1 and Type 2 diabetes) drawn from two ethnic populations, including 471 patients with nephropathy and 494 control diabetic patients without nephropathy. A C/T transition at position -106, and a (CA)n microsatellite marker 2.1 kb from the start site of the aldose reductase gene were genotyped in nephropathic patients and non-nephropathic controls from each cohort. RESULTS:Carriage of the -106 T allele was significantly associated with diabetic nephropathy in three of the four study groups. The Mantel-Haenszel combined odds ratio was 2.22 (95% CI 1.69, 2.94), P = 1.05 x 10(-8). We found no evidence for association of the microsatellite marker with nephropathy, despite moderate levels of disequilibrium between the two markers. Meta-analysis of published data yielded no evidence for association of the microsatellite marker with diabetic nephropathy in Type 2 diabetes, but varying degrees of association with diabetic nephropathy in Type 1 diabetes. CONCLUSIONS:Meta-analyses provide more convincing evidence of a role for the ALR2-106 marker than for the microsatellite marker in diabetic nephropathy (DN). More studies are now required to confirm these results and to establish whether the ALR2-106 polymorphism has a functional role in DN.