Interest in health status indicators has produced measures of widely varying applicability. We were interested in the use of such indices to establish mean recovery curves for groups of similar patients undergoing acute hospital treatment. A longitudinal study relating resource usage to these recovery curves had been intended but there were difficulties in finding suitable indicators. Firstly, two published indicators relying on patient interviews were tested for consistency. Poor correlations were found among those scorers unfamiliar with the patients and it seems unlikely that these indicators could be used in a routine system. Different parts of the indices presented difficulties to the different professions involved in scoring, and a multidisciplinary approach may be needed in assessing full health. The indicators tested included no assessment of prognosis. Those parts of the indices which had produced significant correlations were retained in subsequent work and were supplemented by further measures designed to overcome the earlier difficulties. A new trial of this indicator was undertaken where staff, familiar with the patients, scored data recorded by the Problem Oriented Medical Record system. This produced improved correlations but some problems remain.