Circulating anodic antigen (CAA) levels in serum and faecal egg counts are both quantitative measures of Schistosoma mansoni worm burdens. In this study, we have tested whether circulating anodic antigens can be included into an established egg count model. A data set with 3 repeated faecal egg count and serum CAA measurements of 50 individuals from a community in Burundi with moderate endemicity was used. By means of Monte Carlo simulation, both antigens and egg counts were related to an underlying worm pair distribution, taking into account the variation in repeated measurements (within individuals) and the variation in worm burdens (between individuals). Models with various assumptions (e.g. presence or absence of density-dependent egg production) were tested. Whereas observed and predicted egg counts agreed fairly well, the circulating antigen data could not be described satisfactorily. In particular, the predicted number of negative antigen concentrations was much lower than observed, while the number of light positives was overestimated. There seems to be a mechanism that causes a shift of expected (low) positive CAA concentrations towards zeros, which the proposed models do not provide for. Possible biological as well as assay-related mechanisms that may account for this shift are discussed. The assumption that serum CAA concentrations are a simple direct reflection of worm (pair) burdens could not be corroborated by this modelling exercise. Apparently, the relationship between (measured) CAA concentrations, egg counts and worm burdens in human S. mansoni infections is more complex than assumed.