Abstract In France, bovine brucellosis is subject to stringent surveillance to ensure early detection of re-emergence and maintain the country's disease-free status. However, the cost-efficiency of such surveillance systems has rarely been assessed. This study aimed to: (1) evaluate the detection delay of bovine brucellosis for current and alternative surveillance systems following its reintroduction; and (2) estimate the associated surveillance and control costs. A mechanistic, stochastic, agent-based simulation model was developed to represent the spread of bovine brucellosis within and between cattle farms in mainland France after the introduction of an infected animal. The results showed: (1) improved efficiency (considering both effectiveness and cost) when suspending testing at introduction; (2) reduced efficiency when annually screening all adults in one-third of suckler herds compared to 20% of adults in all suckler herds; and (3) slightly improved efficiency when reporting abortion series instead of all individual cases. The current surveillance system detected infection within a median delay of 49–51 weeks, with a median of one infected farm at confirmation, regardless of the epidemiological context. For alternative systems, detection occurred within a median of 40–99 weeks, with one to three infected farms. These findings suggest that, under model assumptions, all surveillance systems maintained a relatively low number of infected herds. Our study provides cost-effectiveness information that will help stakeholders compare the efficiency of alternative scenarios, improve the current surveillance system and make relevant decisions about the allocations of resources.