This study was conducted for the purpose of determining the frequency of medica¬tion errors (MEs) occurring in a teaching psychiatric hospital in the city of Tehran, Iran. Medication errors were defined by using widely accepted criteria. A cross-sectional prospective study using chart reviews to detect medication errors. Rates of error in prescribing, ordering, transcribing, administering and monitoring were determined. The frequency of these errors was analysed and reported using SPSS-21 software. The study was conducted on six patient care units (n=182). We followed patients for two weeks from the first day of admission in any of the six units. All of 20674 doses were studied in the wards in order to detect prescribing, ordering, transcribing, administrating and monitoring errors. In chart review, we detected a total of 1375 errors in 20674 opportunities for errors (6.7%). In each stage, the frequency of medication errors was: Prescribing: 2.4 %, Ordering: 12.5 %, Transcribing: 3.7 %, Administration: 81.1 %, and finally Monitoring: 0.3 %. The most common types of error throughout the medication process were: wrong dose, omission of dose, unordered dose. There is a need for quality improvement as almost 50% of all errors in the medication process were caused by missing actions. We assume that the number of errors could be reduced by simple changes to existing procedures or by implementing automated technologies in the medication process. Clear guidelines must be written and executed to reduce the incidence of medication errors.