Objectives: In France, cancer registries cover only 20% of the population. To estimate regional cancer incidence (I), various indicators can be used: mortality (M), hospital stays reported by the Program for the Medicalization of Information Systems (PMSI), and Long-Term Conditions (ALD). The objective of this study was to evaluate the validity of these surrogates and to rank them to provide regional incidence estimates for the 2008–2010 period. Method: The estimation principle is identical for all three methods and consists of applying the M/I ratio from the registry area (or ALD/I or PMSI/I, respectively) to regional mortality data (or ALD or PMSI, respectively). Due to the departmental coverage of registries in France, these methods can only be evaluated—and thus validated—at this administrative level in departments covered by a registry. A decision tree was used to classify each method by cancer type and then rank them. Eligible sites (where at least one method was deemed valid) were identified. However, ranges of estimates were provided for ineligible sites. Results: Among men, 8 out of 19 sites were eligible, including 6 based on ALD and 2 on PMSI. Among women, 9 out of 21 sites were eligible, including 3 based on ALD, 5 on PMSI, and 1 on mortality. Conclusion: This study presents the first structured effort to evaluate different approaches to estimating regional incidence. It confirms the value and limitations of medical-administrative databases for this issue. (R.A.)