The natural contamination of foods with a bacterial pathogen frequently consists of a mixture of strains with their own characteristics of survival, growth potential and virulence. Quantitative Microbial Risk Assessment (QMRA) must account for this genetic diversity to reflect the variability of the pathogen risk and to identify the genetic groups present at key stages of the food pathway. To describe the transmission dynamics of a heterogeneous population of B. cereus, we developed an exposure model that covers a food processing chain from “farm to table”. The studied food was a cooked, pasteurised and chilled courgette purée used as an example of Refrigerated Processed Food of Extended Durability (REPFED). The B. cereus population consists of a continuum of genetic groups ranging from mesophilic and highly heat resistant, to psychrotrophic and moderately-heat resistant ones. At each step in a processing chain comprising cooking, blending, mixing with ingredients providing a secondary contamination, pasteurisation and chilling for several weeks, the prevalence of contaminated units (batches or packages) and the spore load within the units was determined for each genetic group, as well as their proportion to the total B. cereus population in the units. The model predicted that all packages contain mesophilic groups just after partitioning. The addition of mesophilic strains by the ingredients during the process of the courgette purée was an important contribution. At the end of the domestic storage, the model predicted a dominance of the mesophilic groups, while only some psychrotrophic groups were present