The Government of Tamil Nadu state in India has been implementing various health sector reforms (for example, expansion and upgradation of public health facilities, provision of round the clock services in selected primary health centres and continuous availability of quality medicines decentralisation) in a bid to improve efficiency in health care. However, few attempts have been made to make an estimate of the efficiency of hospitals in Tamil Nadu as well as in India till date. The objectives of this study are: (i) to estimate the relative technical efficiency (TE) and scale efficiency (SE) of a sample of public hospitals in Tamil Nadu; and (ii) to demonstrate policy implications for health sector policy makers. The Data Envelopment Analysis (DEA) approach, a well-known operations research (OR) technique for evaluating the relative efficiency of a set of similar decision making units (DMU), was used to estimate the efficiency of these hospitals. To do so we made use of the data collected from the Directorate of Medical and Rural Health Services (DMRHS) for 29 districts of Tamil Nadu in 2004–05. The output data included are outpatient visits, number of inpatients, number of surgeries undertaken, number of deliveries and number of emergency cases. The numbers of staff members and bed strength were used as input. Of the 29 hospitals, it was found that 52 per cent were technically efficient as they had relative efficiency score 1.00 and lie on the efficiency frontier, while the remaining 48 per cent were technically inefficient and can use some of the efficient hospitals as their peers to improve their efficiency. Further, the average scale efficiency among the inefficient hospitals was 81 per cent, which implies that the scale inefficient hospitals could reduce their size by 19 per cent without reducing their current output levels.