1. We have retrospectively analysed data collected by a local adverse drug reactions reporting scheme in an acute hospital medical setting and have determined the numbers and types of reactions that would have merited notification as yellow card reports according to the guidelines of the Committee on Safety of Medicines. 2. The data related to 20,695 consecutive acute general medical admissions on seven general medical wards (140 beds) and were collected over 3 years, from April 1990 to March 1993. 3. Over 3 years there were 1420 reports of suspected adverse drug reactions, a rate of 68.7 per 1000 admissions. 4. If the guidelines for reporting issued by the Committee on Safety of Medicines had been strictly followed, 477 yellow cards would have been sent (23.1 per 1000 admissions). In 357 of these reports (74.8%), the reaction had caused admission to hospital. Only 31 of the 477 potential cards (6.5%) involved black triangle drugs and 10 of these were for minor reactions. 5. Only 30 of the 477 potential yellow cards (6.3%) were known to have been sent. The majority of those reactions not reported were for drug-related admissions, most of which were for well-known reactions to established drugs. 6. We have confirmed and quantified the extent of under-reporting of serious suspected adverse drug reactions to the Committee on Safety of Medicines from our hospital medical unit.