Background: Voluntary hospital reporting systems are potentially valuable sources of information about medical errors and adverse events. This study examined the extent and variation in the reporting of medication errors and adverse drug events in a voluntary hospital incident reporting system. Methods: A retrospective analysis of received incident reports of potential and preventable adverse drug events over a 22-month period was conducted at a 1,300-bed, university-affiliated, tertiary care hospital. Reporting of adverse drug events into the hospital's online Risk Management Event/Incident Entry System (RMEES), which is mainly used by nurses, was compared to reporting by pharmacists into a pharmacy reporting system (PHRED). Results: During the study period, the reported preventable and potential adverse drug event rates were 0.47 and 1.85 per 1,000 patient days, respectively, for RMEES-reported events compared with rates of 0.08 and 41.5 per 1,000 patient days for PHRED-reported events. Significant differences by service (P < 0.001) were present for potential adverse drug event rates in both RMEES and PHRED, but preventable adverse drug event rates did not differ significantly (P > 0.05) by service. A modest relationship (R2 = 0.27) between potential and preventable adverse drug event rates reported to RMEES was present. The median ratio of potential to preventable adverse drug events in RMEES was 4.5 (range = 0 to 16). The median ratio of PHRED to RMEES reports was 7.8, but varied markedly among individual nursing units (range = 0 to 157). Conclusions: Compared with rates reported in the literature, voluntary incident reporting yielded a much lower reporting rate of adverse drug events with considerable variation in reporting among units and service areas. Voluntary reporting of medical errors and adverse events is unlikely to yield reliable estimates of event rates.