Altogether twenty-six elderly subjects (aged 65-74 years) with persistent impaired glucose tolerance (World Health Organization (1985) criteria) identified in a population-based study, were randomly treated either with chromium-rich yeast (160 micrograms Cr/d) or with placebo for 6 months. The 24 h urinary Cr increased from 0.13 (SE 0.03) to 0.40 (SE 0.06) micrograms/d in the Cr group (n 13) but no change was found in the placebo group (n 11) (0.13 (SE 0.02) v. 0.11 (SE 0.02) micrograms/d). No significant change was observed in the oral glucose tolerance test (glucose dose 75 g; 0, 1 and 2 h blood glucose respectively): 5.3 (SE 0.1), 9.3 (SE 0.3), 8.2 (SE 0.3) mmol/l v. 5.0 (SE 0.1), 8.5 (SE 0.4), 7.3(SE 0.5) mmol/l in the Cr group; 4.9 (SE 0.2), 9.2 (SE 0.6), 8.1 (SE 0.3) mmol/l v. 4.8 (SE 0.2), 8.5 (SE 0.5), 7.0 (SE 0.6) mmol/l in the placebo group (baseline v. 6 months). Glycosylated haemoglobin, plasma insulin, C-peptide and apolipoprotein A1 and B levels remained unchanged, and no improvement was seen in serum total cholesterol (6.2 (SE 0.3) v. 6.4 (SE 0.3) mmol/l for the Cr group, 6.2 (SE 0.4) v. 6.5 (SE 0.3) mmol/l for the placebo group), high-density-lipoprotein-cholesterol (1.1 (SE 0.1) v. 1.2 (SE 0.1) mmol/l for the Cr group, 1.0 (SE 0.1) v. 1.1 (SE 0.1) mmol/l for the placebo group) or triacylglycerols (2.5 (SE 0.4) v. 2.0 (SE 0.4) mmol/l for the Cr group, 2.4 (SE 0.2) v. 2.5 (SE 0.2) mmol/l for the placebo group). The present results indicate that Cr supplementation does not improve glucose tolerance or serum lipid levels in elderly subjects with stable impaired glucose tolerance.