Lead exposure is associated with poor cognitive development in children. Very few studies in sub-Saharan Africa (SSA) have studied blood lead levels (BLLs) and non-gasoline sources of exposure in children. Data from a birth cohort in Benin (2011–2013) suggested that 58% of 1-year-old children had BLLs > 50 ug/L. We aimed to investigate the prevalence of elevated BLLs (>50 µg/L and >100 µg /L) among 425 of these children at 6 years of age in 2016–2018 and to compare BLLs between age 1 and 6 years, and study sources of lead at age 6 years. BLLs were analysed by inductively coupled plasma mass spectrometry. Multiple linear regression and quantile regressions were used to study potential sources of lead. The prevalence of BLLs > 50 µg/L in children was 59.5% (Geometric Mean (GM) 56.4 µg/L, 95% CI: 54.1–58.7) at 6 years of age compared to 54.8% (GM 56.5 µg/L, 95% CI: 53.4–59.6) at 1 year of age. The prevalence of children with BLLs > 100 µg/L decreased from 14.4% at 1 year of age to 8.2% at 6 years of age. After adjustment for all other covariates, consumption of peanuts more than once per month was significantly associated with a 22.0% (95% CI: 4.6, 42.5) increment in BLLs at age 6 years compared with no consumption. Consumption of bushmeat killed by lead bullets at age 6 years was associated with an increase in the higher percentiles of BLLs (P75) compared with the absence of this source. Other potential sources of lead associated with BLLs with marginal significance were consumption of rice, paternal occupational exposure, and the presence of activity with the potential use of lead. This prospective cohort confirms the persistently high prevalence of elevated BLLs in children residing in a rural region in the south of Benin, as well as the presence of multiple and continuous sources of lead. These results highlight the need for prevention programs to reduce and eliminate lead exposure in children.