The objective of the study was to develop predictive models of growth and survival of potentially probiotic Lactobacillus casei KN291 bacteria in a fermented soy beverage stored at different temperatures. Non-linear predictive models were developed and they were characterized by proper adjustment and good stochastic properties of the parameters assessed. The models developed satisfactorily describe the behaviour of L. casei KN291 bacteria in the fermented soy beverage and can be used to assess (predict) the storing time-depending bacteria count. The fermented soy beverage stored under the storage conditions studied was characterized by the adequately large count of potentially probiotic L. casei KN291 bacteria (approx. 10⁹ cfu/ml) and this fact can guarantee a pro-health effect when eating a 100 g portion of the product.