In this study, we evaluated whether the high number of reports of serum sickness to cefaclor was present in every country and year, and whether these figures from voluntary reporting facilitated an estimation of the relative risk. A nested case-control study was performed with reports of all suspected adverse reactions (ADR) to cefaclor, amoxicillin and cephalexin in the period 1968-1987, as reported to the WHO Collaborating Center for International Drug Monitoring from the U.S.A., the U.K., Sweden, Canada and Germany. The ADR-reporting odds ratio was defined as the ratio of the odds of the number of ADR-reports of serum sickness to cefaclor and amoxicillin or cephalexin and the odds of similar reports of non-serum sickness to cefaclor and amoxicillin or cephalexin. The ADR-reporting odds ratio adjusted for country, age, gender, origin of the report and year of marketing was 12.4 for cefaclor vs amoxicillin and 18.5 for cefaclor vs cephalexin. In children (< 15 years of age) and in adults (> 15 years of age), the relative risk of developing serum sickness of cefaclor vs amoxicillin was estimated at 13.9 (95% confidence interval (95% CI): 6.0-32.2) and 2.9 (95% CI: 0.9-9.4) respectively in the U.S.A., and at 15.1 (95% CI: 7.2-31.5) and 5.5 (95% CI: 2.0-15.0) respectively in the other four countries together. In this study, the ADR-reporting odds ratio facilitated a valid estimation of the relative risk.