Cross-sectional surveys were conducted in 20 randomly selected streets in Moradabad city and two villages in Moradabad district in North India to determine the association of magnesium with risk of coronary artery disease (CAD). There were 501 rural (270 men and 231 women) and 505 urban (250 men and 255 women) subjects between 50-54 years of age inclusive. The overall prevalence of CAD was three times higher in urban than in rural subjects (11.7 vs. 3.98) and the rates were comparable in both sexes. Dietary intake of magnesium was significantly (P < 0.05) higher in rural subjects in both men (520 +/- 58 vs. 415 +/- 47 mg/d) and women (432 +/- 40 vs. 316 +/- 38 mg/d). Dietary magnesium intake and serum magnesium were inversely correlated with CAD. The odds ratio for dietary magnesium intake indicates a higher prevalence of CAD at lower intakes of magnesium in both rural (0.67, 95 per cent confidence interval 0.51 to 0.86) and urban (0.72, 95 per cent CI 0.54 to 0.90) subjects. Multivariate regression analysis showed that serum and dietary magnesium were significantly associated with CAD. Hypertension was not associated with CAD, and serum cholesterol showed only weak association in both rural and urban subjects. The inverse association of dietary and serum magnesium with CAD shows that some urban populations of India may benefit from increased consumption of dietary magnesium and higher serum magnesium levels.