The growing application of evidence-based medicine practices across U.S. healthcare has created greater dependence on information resources, especially related to quality and consistency of data. The manipulation of data through coding and classification of patient information presents a critical process where the quality of information, as well as perceived quality of care, could potentially suffer. Where recent regulatory standards, such as HIPAA, create additional requirements for consistency in coding of health information, it becomes apparent that meaningful health outcomes assessment is, in part, an indicator of data quality as well as clinical quality. In a national survey of 16,000+ accredited health information managers we found most respondents reported that significant coding errors existed in 5% or less of the records in their institutions. Within specific organizations, however, coding errors existed in six to ten percent of their records, and at times exceeded 20%. Regional variation in reported coding error and inconsistency ranged widely, occurring across organizations as well as population concentrations. Metropolitan-based organizations tended to have somewhat worse reported overall coding accuracy, compared to suburban and rural areas. At a national level there will need to be some degree of coding and classification uniformity across population areas, if healthcare professionals are expected to rely on comparative evidence benchmarks to fully assess medical outcomes data. Related impacts on comparative cost and clinical performance assessment are discussed.