Buoyancy-driven turbulent dispersion in a maturation pond is studied using a combination of field measurements and computational fluid dynamics. Modelling flow in maturation ponds requires turbulent closure models because of the large physical size and the need to model on diurnal timescales. Simulation results are shown to be more sensitive to the inclusion of a buoyancy production term appearing in the turbulent transport equations than to the model choice. Comparisons with experimental thermal profiles show that without this term, thermal mixing is over-predicted. When including the term, stratification occurs but thermal mixing is under-predicted in the lower water column. In terms of pond performance, the effect of this term is to cause increased surface die-off of Escherichia coli during sunlight hours due to the generation of stratification. It is recommended that future modelling consider and implement this term.