Listeria innocua CLIP 20595 was grown on the surface of a gelatine gel submitted to a drying process, that is, under constant air temperature and three controlled air velocity and relative humidity conditions. To predict the growth of L. innocua, a water transfer model and a predictive bacterial model were combined. Both models were first optimised and validated on separate experimental data. Experimentally, L. innocua growth slowed down and maximal population decreased when air velocity increased and relative humidity decreased: the water transfer coefficient at the air/product interface was sensitive to the sample location during drying. The combined water transfer/growth models accurately predicted L. innocua behaviour under the three drying conditions. (c) 2006 Elsevier Ltd. All rights reserved.