Objective Caesarean rate increased in France between 1995 and 2003, but remained stable between 2003 and 2010. Our objective was to analyse these trends by identifying the groups of women who contributed to the increase and those who contributed to the stabilisation. Design Cross‐sectional population‐based study from the French national perinatal surveys. Setting All maternity units in France. Population Representative samples of women delivering in 1995 (n = 13 147), 2003 (n = 14 482), and 2010 (n = 14 681). Methods Robson classification, based on pregnancy and delivery characteristics, was used for each group. Main outcome measures Caesarean rate for each group, its contribution to the overall caesarean rate, and the differences (Δ) in these contributions between 1995 and 2003 and between 2003 and 2010. Results Overall caesarean rates were 15.4% in 1995, 19.7% in 2003 and 20.5% in 2010. Between 1995 and 2003, the contribution to the overall caesarean rate of all groups but one rose. Between 2003 and 2010, the contribution of all groups but three stabilised or decreased: nulliparous women in spontaneous labour with singleton cephalic fetuses at term (Δ = + 0.5%, 95% CI 0.1–0.9%), an increase explained by their higher caesarean rate; nulliparous women with induced labour at term (Δ = + 1.1%, 95% CI 0.8–1.4%) caused by an increase in both the caesarean rate and the relative size of this group; and women with previous caesarean (Δ = + 0.8%, 95% CI 0.3–1.3%), because of the growing size of this group. Conclusion Proposing and evaluating interventions for improving the management of labour in nulliparous women could help to maintain caesarean rates and mitigate increases among multiparous women in the future.