Renal changes with time have been studied in 14 workers engaged in the production of cadmium (Cd) in a zinc ore refinery. These workers were examined once a year in the period 1980 to 1985 and 13 of them also in 1989. Four of the workers (group A) had been employed in an old Cd plant before 1973 and had received higher exposures to Cd than the other workers (group B). Average urinary Cd concentrations over the whole study period in workers of group A ranged from 6.9 to 9.2 micrograms/g creatinine (median 8.4 micrograms/g) and in workers of group B from 0.64 to 7.1 micrograms/g creatinine (median 1.9 micrograms/g). Renal effects were assessed by the determination of urinary N-acetyl-beta-D-glucosaminidase (NAG), beta 2-microglobulin (beta 2-M), retinol binding protein, albumin, total protein, and serum creatinine concentrations and activity. Urinary beta 2-M concentrations in three of four workers of group A were close to or marginally above the upper normal limit during the study period. The beta 2-microglobinuria was not, however, progressive. No values outside normal limits were detected for any of the other renal tests in workers of groups A and B, related to exposure to Cd. Dose-response relations showed that urinary Cd correlated significantly with urinary NAG activity and total protein and beta 2-M. The earliest change induced by Cd was seen for urinary NAG activity within normal limits of NAG excretion. The regression lines were similar in the surveys between 1981 and 1989, indicative of no progression to higher values for any of the renal tests. The current biological exposure index (BEI) of 10 micrograms/g creatinine for workers exposed to Cd, set by the American Conference of Governmental Industrial Hygienists (ACGIH), therefore seems justified, although the safety margin is small. The World Health Organisation recommended limit and ACGIH (1992-3) proposed limit of 5 micrograms/g creatinine would provide a much larger safety margin, and could be regarded as an action point for increased health surveillance.