This study describes the different stages of optimization in an original drying process for lactic acid bacteria that allows the retrieval of dried samples of Lactobacillus plantarum with maximum viability. The process involves the addition of casein powder to bacterial pellets, followed by mixing and then air-drying in a fluidized bed dryer. The effects on bacterial viability of the a(w) of the casein powder and the kinetics of aw variation in the fluidized bed dryer are considered. These parameters were first studied in a water-glycerol solution and the results were then applied to the drying process. Data from the study in liquid medium were reliable in the fluidized drying stage, insofar as optimal viability was achieved for similar dehydration times (16 - 50 min in liquid medium, and 30 min in the fluidized bed dryer). However, when the powder was mixed rapidly with bacteria, the level of destruction differed from that observed in liquid medium. Viability was up to 70% when the a(w) of water -glycerol was 0.55, whereas it was only 2.1% when the a(w) of the casein-bacterial mix was 0.64. The predictive capacity of dehydration in liquid medium is discussed with regard to the permeability of cells to external solutes. The new process allowed 100% survival of L. plantarum after complete drying (final a(w) < 0.2). However, when used for the desiccation of L. bulgaricus, these parameters achieved a viability of less than 10%. (C) 2004 Wiley Periodicals, Inc.