Modification of the molecular weight distribution of commodity polypropylene resins is frequently practised in industry in reactive extrusion operations using peroxide initiated reactions in the melt. Thus, the processing characteristics can be modified in a controlled fashion by narrowing the molecular weight distribution (MWD) of the starting material by means of random chain scissions. This process has attracted considerable interest in recent years and several experimental and modelling studies have been carried out. In this work, a systematic probabilistic study is carried out for the peroxide initiated degradation of polypropylene, and a stochastic model is developed based on a Monte Carlo simulation algorithm. In this modelling framework, a polymer MWD is defined as an assembly of an entire population of reacting species and its state is characterized by the types of molecules and the number of each type of molecules in the system. Chemical reactions are represented by the transitions of the system from one state to another and the temporal behaviour of the MWD takes the form of a Markovian random walk in the N-dimensional space of the molecular populations of the N species. Using this stochastic model, certain approximations previously used are relaxed and the evolution of the entire MWD is mapped out for different experimental conditions. Monte Carlo simulation results show very good agreement with experimental data as well as with predictions from a deterministic model.