The current study aimed to evaluate whether nicotinamide adenine dinucleotide phosphate (NADPH) alteration in erythrocytes from patients with non-insulin-dependent diabetes mellitus (NIDDM) is responsible for the impaired glutathione (GSH) redox status, and to assess if short-term inhibition of the polyol pathway normalizes NADPH levels and GSH redox status via an amelioration of the NADPH/total NADP (tNADP) ratio. For this purpose, erythrocyte NADPH and GSH levels were measured in 18 NIDDM patients at baseline and then after 1 week of random double-blind assignment to treatment with either tolrestat (an aldose reductase inhibitor, 200 mg daily) (n = 12) or placebo (n = 6). A group of 16 healthy volunteers served as the control. In the basal condition, mean GSH (P < .0001) and NADPH (P < .0001) levels and NADPH/tNADP (P < .0001) and GSH/ glutathione disulfide (GSSG) (P < .005) ratios were lower in NIDDM patients than in control subjects. Tolrestat treatment increased GSH levels (P < .05 v placebo and baseline) and the NADPH/tNADP ratio (P < .05 v placebo and baseline). Interestingly, tolrestat-induced changes in GSH and NADPH levels and in GSH/GSSG and NADPH/tNADP ratios were significant only in patients who showed a decreased NADPH/tNADP ratio at baseline (n = 8). In these latter patients, we also found a direct correlation between percentage increments in GSH levels and NADPH/tNADP ratios after tolrestat treatment (r = .71, P < .05). In conclusion, our findings support the hypothesis that polyol pathway activation decreases NADPH and GSH levels. Accordingly, short-term inhibition of this enzymatic route increased both the GSH level and the NADPH/tNADP ratio. These changes were observable only in the subgroup of patients with an abnormal NADPH/tNADP ratio at baseline. Polyol pathway inhibition could be useful for decreasing oxidative stress in NIDDM.