This study assessed the possibility of relationship between manganese and lead levels in blood in schoolchildren residing in different geographical regions of South Africa. A cross-sectional survey was carried out in schools of three cities: Cape Town (11 schools), Johannesburg (10 schools), Kimberley (six schools) and in the Northern Cape in three rural sites (four schools). A total of 1282 venous blood samples were collected from grade one children. The relationships between blood manganese and blood lead levels (treating each in turn as the response variable in order to adjust for the effect of confounding variables) were investigated by fitting mixed models. Mixed models were fitted with natural log (manganese concentration) as the response variable, and blood lead level as the principal explanatory factor. The model also included terms for centre and a centre by lead interaction and examined potential confounders. The important confounders were found to be gender, race and whether there was paint peeling from the outside walls. There was a significant centre by lead interaction (P<0.0001) with the effect of lead on ln(Mn) being different in the various centres. Mixed models fitted with blood lead level as the response variable and blood manganese as the principal explanatory factor, with terms for centre, a centre by manganese interaction and potential confounders, again found overwhelming evidence (P<0.0001) of a centre by manganese interaction. The study found great variability in both blood lead and manganese levels, both within and between sites, and there was not a consistent relationship between the two metals in the four sites.