The objectives are to investigate the incidence of childhood leukaemia in young people around the French nuclear waste reprocessing plant (La Hague, Normandy) for the period 1978-1992, and to check the sensitivity of the results to the choice of disease clustering tests and reference rates. The performance of three statistical procedures are compared: a conventional approach; a Poisson maximum test, and an extraction mapping technique. On the whole, the results are in general agreement and demonstrate the apparent existence of a distinct cluster of childhood leukaemia, contiguous and south-east of the La Hague nuclear facility.