Background: The inconsistent definition of non-therapeutic medication omissions, underreporting, and a poor understanding of their associated factors hamper efforts to improve medication administration practices. Aim: To examine the incidence of non-therapeutic medication omissions among acutely ill medical and surgical adult patients; and to identify the patient-, drug- and system-related predictors of these omissions. Method: A medication chart audit of 288 acutely ill adult medical and surgical patients admitted to 4 target wards (2 surgical and 2 medical) at an Australian hospital. Patients admitted to these wards from December 2008 to November 2009, with at least one regularly prescribed medication, were eligible. The sample was stratified according to gender, season and ward. A medication chart audit identified medication omissions, and data were collected on gender, age, length of stay, comorbidities, medication history and clinical pharmacy review. Results: Of the 288 medication charts audited, 220 (75%) had one or more medication omissions. Of the 15 020 medication administration episodes, there were 1687 omissions, resulting in an omission rate per medication administration episode of 11%. Analgesics and aperients were the most frequently omitted medications, with failure to sign the medication record and patient refusal, the main reasons for omission. Female gender (p < 0.001) and the number of medication administration episodes (p < 0.001) were statistically significant predictors of non-therapeutic medication omissions. Conclusion: The high incidence of medication omissions suggests there is need for an agreed definition of medication omission and its inclusion as a reportable incident. Increasing medication reconciliation via implementation of the Medication Management Plan may also reduce the opportunity for error.