Aim The aim of this study was to identify factors predictive of serious infections over time in patients with systemic lupus erythematosus (SLE). Methods A multi-ethnic, multi-national Latin American SLE cohort was studied. Serious infection was defined as one that required hospitalization, occurred during a hospitalization or led to death. Potential predictors included were sociodemographic factors, clinical manifestations (per organ involved, lymphopenia and leukopenia, independently) and previous infections at baseline. Disease activity (SLEDAI), damage (SLICC/ACR Damage Index), non-serious infections, glucocorticoids, antimalarials (users and non-users), and immunosuppressive drugs use; the last six variables were examined as time-dependent covariates. Cox regression models were used to evaluate the predictors of serious infections using a backward elimination procedure. Univariable and multivariable analyses were performed. Results Of the 1243 patients included, 1116 (89.8%) were female. The median (interquartile range) age at diagnosis and follow-up time were 27 (20–37) years and 47.8 (17.9–68.6) months, respectively. The incidence rate of serious infections was 3.8 cases per 100 person-years. Antimalarial use (hazard ratio: 0.69; 95% confidence interval (CI): 0.48–0.99; p = 0.0440) was protective, while doses of prednisone >15 and ≤60 mg/day (hazard ratio: 4.18; 95 %CI: 1.69–10.31; p = 0.0019) and >60 mg/day (hazard ratio: 4.71; 95% CI: 1.35–16.49; p = 0.0153), use of methylprednisolone pulses (hazard ratio: 1.53; 95% CI: 1.10–2.13; p = 0.0124), increase in disease activity (hazard ratio: 1.03; 95% CI: 1.01–1.04; p = 0.0016) and damage accrual (hazard ratio: 1.22; 95% CI: 1.11–1.34; p < 0.0001) were predictive factors of serious infections. Conclusions Over time, prednisone doses higher than 15 mg/day, use of methylprednisolone pulses, increase in disease activity and damage accrual were predictive of infections, whereas antimalarial use was protective against them in SLE patients.