A diffusion controlled precipitation model based on classical nucleation and growth theory has been implemented to simulate the precipitation kinetics in a hot rolled supersaturated Al-Mn alloy (AA3103). The modelling approach explicitly includes the effect of concurrent recrystallisation on precipitation and considers the simultaneous evolution and interaction of two precipitate populations that vary significantly in size, i.e. constituent particles and dispersoids. Comparison with experimental results shows that this classical modelling approach predicts incorrect nucleus density and too high precipitation rates, which cannot be simply corrected by parameter fitting. Reasons for this discrepancy are discussed in terms of selection of nucleation sites, the effect of diffusion in a multi-component system, various diffusion paths and the possible influence of precipitate shape and size distribution. The model is subsequently altered by introducing two additional parameters that control the Mn solute concentration at the particle-matrix interface. This more phenomenological model is successful in reproducing the experimental precipitation kinetics, both in deformed and undeformed aluminium matrix, and the effect of concurrent recrystallisation for a wide temperature range.