The growth and survival behaviours of Bifidobacterium lactis and Lactobacillus acidophilus in a semi-hard Gouda cheese at various axial locations during 9 wk of ripening at 13 degrees C were assessed using non-linear regression analysis. The final average salt levels ranged in 2-4%(w/w). Viable numbers of both probiotic strains underwent a slow decline during the first 3 wk followed by a sharper decrease towards the end of ripening; such decrease was more substantial for the outer than for the inner cheese portions. Salt transport was successfully described by Fick＇s second law of diffusion, and the cheese was considered as a finite slab for modelling purposes. Salt diffusivity remained constant with time and was estimated to be 0.2 cm(2)/day. Theoretical salt concentration profiles were in good agreement with experimental data. The mathematical models postulated and fitted to the microbial viability data encompassed both a linear relationship between specific death rate and salt concentration and a constant death rate, following a methodology of increasing model complexity. Decision on the better model was taken based on a F-test of the ratio of incremental sum of squares of residuals to sum of squares of residuals of the more complex model, and it was concluded that viability of the probiotic strains was better described by a first order process independent of local salt concentration. Prediction of profiles of viable numbers of B. lactis and L. acidophilus in cheese with respect to both ripening time and axial location for several overall salt concentrations is useful in attempts to predict potential viability by the time of consumption.