BACKGROUND: In this era of declining cardiovascular mortality, other measures of Quality Assurance (QA) become important. We were concerned that the Discharge Abstract Database (DAD) provided through Hospital Records was neither timely nor complete. OBJECTIVES: To analyze incongruent information, determine sources of error, and provide a true incidence of complications. METHODS: Five cardiologists used Palm OS based software (Smart-list to Go) to enter all procedures (cardiac catheterization (CC), closed-heart operation (CHO) or open-heart operation (OHO)) and associated complications (chosen from pre-defined pick lists) noted during their on-call duties. Following a brief trial period in 2005, all data captured over 2 years were compared. RESULTS: DAD identified 152 cases not in the Cardiovascular Complications Database (CCD): 53% were out of scope of CCD, 37% missed CC, 4% CHO and 6% OHO. Detailed comparison of 55 OHO cases for all potential complications revealed 46 positive in DAD and CCD, 324 negative in both, 75 positive in DAD, but not in CCD and 50 positive in CCD and not in DAD. This gave a 25% discrepancy between the two systems, accounted for by the use of non-specific coding terms in DAD and a low threshold for recording events. Using the corrected data from CCD, control charts with quarterly rates for all complications were derived: Those related to CC were 1/10 cases, CHO 3/10 cases and OHO 5.5/10 cases. CONCLUSIONS: Significant anomalies were found between the CCD and DAD; creation of reliable QA data requires rigorous protocols congruent between physician and administrative perspectives.