Outcome measures have been developed in an acute hospital for specific patient groups (primarily cholecystectomy, diabetes, coronary angioplasty and knee replacements). For each condition a set of indicators was derived which ranged from clinical and laboratory measures to measures of general health status. All indicators attempt to show changes in patient health over time. It has been shown possible to collect the necessary data for such outcomes measures. The costs and methods of data collection varied between conditions. Patient completed questionnaires were found to be particularly useful and in inpatient studies have given high response rates (over 95%) for postal follow-ups and have been validated by interviews. The differing ability of the various indicators to show clinical changes has been demonstrated. In all specialties there was found to be generally high levels of association between different indicators. The information collected on patient outcomes was presented at meetings of the various clinical teams and the value of the information in promoting practical change was examined. It was concluded that different indicators have different value in such reviews and that three key characteristics are identified. The first concerns whether the measures reflect clinical or patient's perceptions of health. The second concerns the extent to which an indicator is a direct measure of health or a proxy (or process) measure. The third concerns the extent to which an observed outcome indicator can be linked to particular processes of care. The study has generated support from the clinicians and it is suggested has changed their views on how they judge their own performance. In some instances practical changes have resulted form the presence of the outcome information. The potential future roles for outcome measurement in the health service is discussed.