BACKGROUND: Low medication persistence is reported in patients with severe hypertension but few data are available according to drug classes. METHOD: Adults without cardiovascular disease who started treatment, in a semester between 2010 and 2012, with two concurrent dispensings of at least three classes were identified in the French national health data system. High persistence after 12 semesters of follow-up was defined by a 6-monthly mean of number of drug classes equal to or greater than 80%. RESULTS: Five hundred and seventy-six thousand and forty-eight adults alive at 6 years were included (three classes: 79%, four classes: 18%, five or more classes: 4%) with a mean age of 65.3 years, constituting 42% of men. High persistence was observed for 72% of people and multivariate analysis of baseline factors found a negative association for: female sex, extreme ages, living in an overseas department, at least one comorbidity, absence or frequent general practitioner consultations and a cardiologist consultation. The adjusted odds ratio was low for dispensing of ACEIs (0.87; 95% CI: 0.8-0.95), other RAS antagonists (0.91; 95% CI: 0.83-0.99) and a high number of classes (4: 0.17; 95% CI: 0.15-0.19, five and more: 0.06; 95% CI: 0.05-0.08). An inverse association was observed for diuretics (1.45; 95% CI: 1.33-1.59), calcium channel blockers (1.63; 95% CI: 1.50-1.79), beta-blockers (1.92; 95% CI: 1.76-2.1) and other antihypertensive classes (1.6; 95% CI: 1.5-1.8). No significant association was observed for ARBs (1.0; 95% CI: 0.9-1.1). CONCLUSION: These results based on a large primary prevention population should encourage the implementation of new pharmacological and nonpharmacological management strategies for people with severe hypertension in France.