Introduction: Leprosy reactions, the main cause of neural damage, can occur up to 7 years after starting multidrug therapy. We aimed to approach the prognostic factors that may influence the leprosy reactions over the follow-up time. Methods: Retrospective cohort study, encompassing 10 years of data collection, composed of 390 patients, divided into 201 affected by reactions and 189 reaction-free individuals. Epidemiological, clinical, and laboratory variables were approached as prognostic factors associated with leprosy reactions. The association among variables was analyzed by a binomial test and survival curves were compared by the Kaplan-Meier and Cox proportional-hazards regression. Results: 51.5% (201/390) of patients were affected by leprosy reactions. These immunological events were associated with lepromatous leprosy (16.2%; 63/390; p < 0.0001) and multibacillary group (43%; 169/390; p < 0.0001). This study showed that survival curves for the prognostic factor anti-PGL-I, comparing positive and negative cases at diagnosis, differed in relation to the follow-up time (Log Rank: p = 0.0760; Breslow: p = 0.0090; Tarone-Ware: p = 0.0110). The median survival times (time at which 50% of patients were affected by leprosy reactions) were 5 and 9 months for those reactional cases with negative (26/51) and positive serology (75/150), respectively. The time-dependent covariates in the cox proportional-hazards regression showed anti-PGL-I as the main prognostic factor to predict leprosy reactions (hazard ratio=1.91; p = 0.0110) throughout the follow-up time. Conclusions: Finally, these findings demonstrated that anti-PGL-I serology at diagnosis is the most important prognostic factor for leprosy reactions after starting multidrug therapy, thus enabling prediction of this immunological event.