Data envelopment analysis (DEA) was used to investigate the efficiency of a set of district hospitals in the state of Tamil Nadu. These facilities naturally provide primary and secondary care, but are also expected to function as health centres, addressing mostly preventive medicine, hygiene and other public health issues. This study aimed to obtain insight on their technical efficiency in light of their particular role. We have considered all the 29 district hospitals in the state. Variables chosen to characterise production were numbers of assistant surgeons, civil surgeons, staff nurses and beds as inputs; and in-patient and outpatient visits, surgeries both major and minor, and deliveries performed as outputs. The DEA model was input-oriented, allowed for variable returns to scale, and units were ranked according to a benchmarking approach. Results indicate that eight of the 29 hospitals (27 per cent) are efficient, the DEA score being 1.0, while the remaining 21 hospitals (72 per cent) are relatively inefficient, needing to benchmark their performance with that of their peer group.