Objective: To identify, classify and determine the factors associated with medication discrepancies for type 2 diabetes mellitus (T2DM) patients, referred from primary care to a tertiary ambulatory clinic. Design: Retrospective audit of outpatient clinic records. Setting: Royal Prince Alfred Hospital (RPAH) Diabetes Ambulatory Care Centre. Participants: 300 randomly selected adult T2DM patients who attended the Diabetes Centre between 01 January 2010 and 31 December 2011. Main Outcome Measures: The rates and types of medication discrepancies were identified by comparing the structured nurse- patient interview (SNPI) with the primary care [General Practitioner (GP)] referral letter, where the SNPI was considered the best possible medication history. Discrepancies were identified as addition, omission, dose and insulin-type discrepancies. Each category was mutually exclusive. Results: Over 80% of referral letters contained at least one discrepancy with a median of two discrepancies per referral. Of a total of 744 discrepancies, the majority were omissions (58.9%). Insulins had the highest discrepancy rate. Factors independently associated with medication discrepancies were GP referral letter type, total number of medications and medication regimen type. Conclusions: A high rate of medication discrepancies was found in GP referral letters for patients referred to this clinic. Automated GP referral letters and inaccurate GP records may have contributed to this, highlighting the need for routine medication reconciliation at transitions of care, to ensure prescribers have access to correct medication information to inform decisionmaking and ensure optimal patient outcomes.