1. In total 1999 consecutive admissions to six medical wards were subjected to a prospective high-intensity drug event monitoring scheme to assess the extent and pattern of admissions caused by adverse drug reactions (ADRs) or dose related therapeutic failures (TF), in a population-based design. The wards were sub-specialised in general medicine, geriatrics, endocrinology, cardiology, respiratory medicine and gastroenterology. 2. Considering definite, probable and possible drug events, the prevalence of drug related hospital admissions was 11.4% of which 8.4% were caused by ADRs and 3.0% by TFs. There were large inter-department differences. 3. The six classes of drugs most frequently involved in admissions caused by ADRs were anti-rheumatics and analgesics (27%), cardiovascular drugs (23%), psychotropic drugs (14%), anti-diabetics (12%), antibiotics (7%), and corticosteroids (5%). Noncompliance accounted for 66% of the TFs with diuretics and anti-asthmatics most frequently involved. 4. The pattern of drugs involved in ADRs was compared with the regional drug sales statistics. Drugs with a particularly high rate of ADR related admissions per unit dispensed were nitrofurantoin and insulin (617 and 182 admissions per 1,000,000 defined daily doses), while low rates were seen for diuretics and benzodiazepines (10 and 7 admissions per 1,000,000 defined daily doses). Confidence intervals were wide. 5. Patients who had their therapy prescribed by a hospital doctor had a slightly higher prevalence of drug events than those who were treated by a general practitioner (12.6% vs 11.8%). The reverse applied for drug events assessed as avoidable (3.3% vs 4.6%). Although these differences were not statistically significant, it may suggest general practitioners as the appropriate target for interventive measures.(ABSTRACT TRUNCATED AT 250 WORDS)