The collection of accurate maternal outcome data enables causes of morbidity and mortality to be identified, which in turn permits interventions to be targeted appropriately. It also allows estimates to be made about the importance of various indicators in predicting birth outcome. These indicators can then be compared between health services, across time and against programme objectives, thus ensuring a management information system that informs policy and provides for real change. A review was done of data collection at the antenatal clinic and maternity ward in a remote rural hospital in northern Ghana. The data collected came from maternity ward records and participant observation, and they highlight deficiencies in the record management procedures. It is argued that exhorting staff to greater accuracy, although obvious, may not be the only solution, because of the structural impediments that often give an illusion of accuracy. The best data need to be collected within the constraints of the equipment and the people. Furthermore, to make the task more meaningful, regular feedback needs to be provided so that the process of record keeping is relevant to those who do it. Ministries of health need to conduct regular audits, like this microanalysis, so that policies are not based on data that are analyzed under a presumption of accuracy.