An analytical mean field model of recrystallization by strain induced grain boundary migration (SIBM), based on cellular growth, has been implemented and compared to experimental measurements of a single-phase Al–0.1Mg alloy deformed to moderate strains. The model is able to predict the onset (nucleation) of primary recrystallization and the kinetics and grain size. Inputs to the model are the prior grain size and the subgrain parameters of the deformation microstructure, together with published values of boundary energies and mobilities. No other parameters are required. The predictions of grain size and recrystallization kinetics are found to be within ~25% of the measured values, and the effects of the limitations of the input data on the model are discussed. The work highlights the difficulties involved with implementing even the simplest physically based model of recrystallization, which are primarily related to problems in adequately characterizing the deformed state and inadequate knowledge of the energies and mobilities of boundaries. Improvements to the model including the incorporation of orientation (texture) dependent stored energies and crystallographic and geometric factors effecting boundary growth rates are discussed.