To quantitate probable adverse drug reactions (ADRs) in a geriatric nursing homes population. A repeated measures prospective study. Two nursing home populations in rural Georgia. All 332 residents present for 30 or more days over a 4-year period. Admission and monthly drug regimen review for each resident, Naranjo algorithm assessment of each ADR, with monthly reports to attending physicians and follow-up within the next month. There were 444 probable ADRs in 217 of 332 residents (67.4%) during this period. The 217 residents had a mean 1.9 +/- 1.3 probable adverse drug reactions (range, 1-9). The ADR group differed statistically from the rest of the population only in the number of drugs per patient (7.8 +/- 2.6 vs 3.3 +/- 1.3), which was almost twice the number of active problems present in both the ADR (4.0 +/- 0.9) and non-ADR populations (3.8 +/- 1.4). The organ systems most commonly involved in the 444 ADRs observed were cardiovascular (188), central nervous system (129), gastrointestinal (82), endocrine (41), immune (17), hematologic (7), pulmonary (6), and renal (5). The drugs most commonly implicated in ADRs were, in decreasing order, diuretics, antipsychotics, anxiolytics, potassium supplements, digoxin, NSAIDs, insulin, theophylline, H2-receptor antagonists, antiinfectives, anticonvulsants, and thyroid supplements. There were 39 multiple drug ADRs in 34 patients. In decreasing order the drug classes in multiple ADRs were CNS depressants, antihypertensives, potassium-altering therapy, and NSAIDs. Numerous patients had repetitions of the same ADR, especially with antipsychotics, NSAIDs, and insulin. ADRs are a common occurrence in a geriatric nursing home population, and may be related to inadequate attention to the patients history as well as to unrealistic therapeutic endpoints.