In France, cancer incidence is measured by departmental cancer registries, which cover one-fifth of the population, while cancer mortality is measured nationwide. National estimates are derived from the ratio of incidence to mortality observed in departments with registries. Thyroid cancer is rarely fatal, and its incidence is highly sensitive to medical practices. Consequently, national estimates for this cancer are imprecise, and this method cannot be applied at the departmental level. The objective of this study was to evaluate the use of hospital data to estimate the incidence of thyroid cancer at the national and departmental levels. It focused on hospital data from the Medical Information Systems Program (PMSI) from 2002 to 2008 and on registry data from 2002 to the most recent year available (2005). The ratio of the number of patients undergoing surgery (PMSI) to the number of new cases of thyroid cancer in departments with a registry (2002–2005) showed significant geographic variability. Consequently, this ratio could not be used to estimate the departmental incidence. At the national level, the ratio allowed for a more precise estimation of national incidence over the 2002–2005 period than using the ratio of incidence to mortality. However, it could not be extrapolated to subsequent years due to a financial revaluation of thyroid cancer procedures in the PMSI starting in 2006. Changes in the annual rates of patients undergoing surgery suggest, despite the limitations of interpretation, a slowdown in the growth of thyroid cancer incidence in certain age groups, particularly between 2002 and 2005. This slowdown would have occurred after 20 years of sustained growth in incidence. It is possible that the effects of medical practices have plateaued or that changes in these practices have begun in certain age groups. However, the recent and short-term trends for this cancer in France remain highly uncertain, and national monitoring of recent trends using the PMSI must continue. (R.A.)