The objectives of this study are: (1) to evaluate, for 24 cancer sites, the validity of departmental estimates of cancer incidence based on data from long-term illnesses (ALD) and registries; (2) to produce departmental estimates for the selected cancer sites. The estimation method uses the ratio of the number of ALD applications to the number of observed incident cases in departments covered by a registry (ALD/incidence ratio). ALD admission data are drawn from the general, agricultural, and self-employed health insurance schemes, while incidence data come from cancer registries. The study period is 2000–2005. The first stage of the evaluation is conducted in departments with registries. The evaluation is based, on the one hand, on criteria for the quality of the estimates and, on the other hand, on a comparison of the quality of the estimates with the extent of geographical disparities in incidence. The quality of the estimates is assessed by examining the geographic variability of the ALD-to-incidence ratio. In the second stage, estimates are produced for the selected cancer sites, based on ALD data from each department and the ALD-to-incidence ratio observed in departments with registries. Eight cancer sites were selected for a departmental incidence estimate: lips-mouth-pharynx, esophagus (men), stomach, colon-rectum, lung, breast, ovary, and testis. The other sites were excluded due to significant geographic variability in the ALD/incidence ratio (liver, gallbladder, pancreas, larynx, melanoma, thyroid, multiple myeloma, leukemia) or because they were too large compared to geographical variations in incidence (cervix, uterine body, prostate, kidney, central nervous system). Finally, three sites were excluded due to specific coding difficulties (bladder) or insufficient sample sizes (esophagus/women, Hodgkin’s disease). Estimates are presented for the eight selected sites. This study presents, for the first time, a systematic evaluation of the validity of departmental estimates of cancer incidence based on ALD data, for 24 sites. This study is a first step. The long-term goal is to define a strategy for estimating the departmental and regional incidence of cancers, based on available medical-administrative and mortality data, for as many cancer sites as possible. (R.A.)