Background: Medication errors can result in harm, unless barriers to prevent them are present. Drug administration errors are less likely to be prevented, because they occur in the last stage of the drug distribution process. This is especially the case in non-alert patients, as patients often form the final barrier to prevention of errors. Therefore, a study was set up aimed at identifying the frequency of drug administration errors and determinants for these errors in an institution for individuals with intellectual disability (ID). Methods: This observational study ("disguised observation") was conducted within an institution in the Netherlands caring for 2500 individuals with ID and lasted from October to December 2004 with a case control design for identifying determinants for errors. The institution consists of both day care units and living units (providing full-time care), located in different towns. For the study, five units from different towns were selected. Within each study unit, the administration of drugs to patients was observed for 2 weeks. In total, 953 drug administrations to 46 patients (25 male, mean age 25.8 years, range 2-73 years) were observed. Results: With inclusion of wrong time errors, 242 administrations with at least one error were observed [frequency = 242/953 (25.4%)] and with exclusion 213 administrations with at least one error were observed [frequency = 213/953 (22.4%)]. Determinants associated with errors were routes of administration "oral by feeding tube" (OR 189.66; 95% CI 46.16-779.24) and "inhalation" (OR 9.98; 95% CI 4.78-20.80), the units "adult full-time care" (OR 2.12; 95% CI 1.05-4.35) and "children daytime care" (OR 10.80; 95% CI 4.43-26.29) and the absence of a distribution robot (OR 4.0; 95% CI 2.67-5.95). None of the identified errors were reported to the voluntary reporting system. Conclusion: This study shows that administration errors in an institution for individuals with ID are common and that they are not formally reported to the voluntary reporting system. Furthermore, it identified some determinants that may be the focus for future improvements aimed to reduce error frequency.