Medication error surveillance is an important component of the quality assurance program of any hospital. The present study was undertaken in an attempt to determine the reliability of data collected retrospectively on medication error forms, as compared to data obtained from a prospective observational study within the same hospital. The results indicate that for many parameters reasonably reliable information can be obtained from voluntarily submitted error reports. Other observations made during the study were also useful to both the nursing and pharmacy departments in their efforts to minimize the occurrence of errors. Based on the information provided by this study, a more formalized medication error surveillance program has been established and a number of changes in the drug distribution and administration systems have been proposed.