The relation of bacteriuria to subsequent mortality was investigated in population-based cohort data from women living in Jamaica and Wales. Three surveys were done in each location over a 13-year period. Women with 10(5) or more gram-negative bacilli per millilitre of urine in three consecutive cultures over a 3-week span were considered bacteriuric for that survey. Mortality was determined at the second and third surveys. Bacteriuria was associated with increased mortality both in the crude data and in life-table analyses adjusted for the confounding effects of age and weight. The adjusted risk ratio for death between the first and third surveys was 1.5 for women bacteriuric at the first survey compared with non-bacteriuric women. For deaths between the second and third surveys, the adjusted risk ratio was highest (2.0) in women who were bacteriuric at both of the first two surveys. The adjusted risk ratio for death between the second and third surveys for women who were bacteriuric at one of the first two surveys but not at the other was 1.6.