While cancer maps are a convenient way of conveying quickly such information as the presence of clusters of areas at high or low risk, the level of statistical significance of these occurrences cannot be easily derived. A method of measuring the "spatial aggregation," based on rank, was presented in the Atlas of Cancer in Scotland (Kemp et al. 1985) and further expanded in the Atlas of Cancer Mortality in Italy (Cislaghi et al. 1986). A refined version of this method is presented and compared with others.