Incompleteness of ascertainment of new cases of any disease leads to underestimation of its incidence rate and may result in false assumptions about incidence trends if incompleteness varies over the course of time (Fraser et al., 1978). We report an attempt to assess the completeness of ascertainment in a cancer registry in the North-Western Region of England using a method based on independent comparison with accurate morbidity data. A total of 1955 verified cases of cancer from 5 independent sources covering 11 sites were used in the study. The corrected mean level of overall registration completeness was found to be 94% but this varied appreciably with site and source of data. Independent comparison is recommended as an effective method for estimating the completeness of cancer registration.