Fermentation temperature (34–40 °C), total inoculum level (1%–3%, v/v) and peeled-fruit-to-water ratio (12.5%–37.5%, w/v) were combined to study their effects on the fermentation of banana media by free and Ca-alginate or κ-carrageenan-immobilised Lactobacillus acidophilus. A three-variable and three-level design method, analysed by response surface methodology (RSM), was used. These factors, except peeled-fruit-to-water ratio, were found to be significantly effective on viable cell numbers and 1-kestose (GF2) concentrations. Contour plots were generated using a graphing software package (Surfer Mapping System, Version 5.0; Golden Software Inc., Golden, CO, USA, 1994) based on fitted quadratic regression equations. Optimum conditions for the highest viable cell number and higher GF2 concentration were found to be around 35 °C fermentation temperature, 1.2% inoculum level and 25.0% peeled-fruit-to-water ratio for Ca-alginate immobilised cell fermentation, and around 39 °C fermentation temperature, 1.8% inoculum level and 25.0% peeled-fruit-to-water ratio for κ-carrageenan immobilised cell fermentation. For free cell fermentation, conditions for the highest viable cell number and higher GF2 concentration could not be obtained. The predicted optimum conditions of immobilised cell fermentation and the experimental values were consistent. This verified the adequacy of these predicted models. It was concluded that RSM was appropriate for the optimisation of banana puree fermentation using cell immobilised L. acidophilus, and products with better synbiotic effects could be obtained. The consumer palatability of the immobilised cell-fermented banana puree was found to be acceptable.