Abstract
This paper details a statistical method for evaluating needle exchange programmes. The approach relies only on needle exchange operations data and the results of HIV tests conducted on needles. We develop statistical models describing the needle infection process and how needle exchange interrupts this process. The method is illustrated using 20 months of data collected in conjunction with the evaluation of New Haven, Connecticut's needle exchange programme, and the results suggest that needle‐borne HIV transmission among participating clients has been reduced by at least 33 per cent.