The effects of different culture parameters and operating strategies were tested on lactic acid production from whey permeate/yeast extract medium by immobilized Lactobacillus helveticus in a continuous two-stage process. High lactic acid productivities of 19-22 g l(-1) h(-1) and low residual sugar was achieved with an overall dilution rate of 0.5 h(-1) and 10 g l(-1) yeast extract. Lowering the yeast extract concentration from 10 to 1 g l(-1) resulted in a gradual loss of activity with time in both reactors, leading to an overall lactic acid productivity of 10.5 g l(-1) h(-1) with 24 g l(-1) residual sugar after 47 h. Inversion of the first and second reactor in the two-stage process led to an important drop in productivity which was only partly restored in the next 3 days of culture. Lactic acid and residual sugar concentrations in both reactors were predicted by a mathematical model, which combined mass transfer equations with a free-cell kinetic model of Lb. helveticus and did not use any fitting factor. Predicted concentrations were close to measured values during the first 18-24h batch or continuous culture and, in general, during pseudo-steady-state continuous cultures in both reactors after this initial colonization period. However, transitions to a new pseudo-steady state, after changes in dilution rate and/or yeast extract concentration, were much slower in the experiment than those predicted by the model. Biomass and pH-gradients within gel beads, measured with fluorescent dyes and microelectrode, respectively, were only qualitatively predicted by the model. Our data suggest that an accurate description of immobilized cell activities cannot be done with kinetic models for free cells, especially for long-culture times, due to a change in the physiology of immobilized cells. (c) 2005 Elsevier Inc. All rights reserved.