The objective of this study was to identify abnormalities in lipid and carbohydrate metabolism in women taking thiazide diuretics and determine whether these abnormalities are mitigated by concurrent postmenopausal estrogen replacement therapy. The study design was cross-sectional; its setting was Rancho Bernardo, an upper middle-class community in southern California. The subjects included 1047 white nondiabetic postmenopausal women, aged 50-89 yr, categorized by the use of thiazide diuretic, estrogen replacement therapy, both, or neither. Medical history including behavior, verified medication use, height, weight, fasting chemistry and lipid panels, and a standardized oral glucose tolerance test with fasting and 2-h plasma glucose and serum insulin levels were determined. Compared with nonusers, women taking thiazides had significantly lower high density lipoprotein cholesterol levels and significantly higher fasting triglyceride, glucose, and insulin levels. Concomitant use of thiazide and estrogen yielded lipid profiles and fasting glucose and insulin levels similar to those of subjects receiving estrogen alone, i.e. elevated high density lipoproteins, decreased low density lipoproteins, and lower levels of fasting glucose and insulin compared with those in nonusers. However, thiazide-associated postchallenge glucose and insulin elevations were not modified by estrogen. These patterns were not explained by differences in age, body mass index, exercise, smoking, alcohol use, type or dose of thiazide diuretic, type of estrogen replacement, or serum potassium levels. We conclude that postmenopausal estrogen use masks thiazide-associated dyslipidemia and fasting elevations in glucose and insulin levels, but does not improve thiazide-associated postchallenge glucose intolerance and hyperinsulinemia. Modification of most of the untoward metabolic effects of thiazides in women taking postmenopausal estrogen could provide a new incentive for the use of this traditional antihypertensive in elderly women.