Microbial lipases (E.C. 3.1.1.3) produced by Geotrichum sp. and Rhizopus sp. were prepared and tested as catalysts in the esterification of isoamyl alcohol and butyric acid in a solvent-free system. Response surface methodology (RSM) based on a five-level, two-variable central composite rotatable design (CCRD) was used to optimise this process, with respect to important reaction variables-substrate molar ratio (RM), temperature and lipase concentration (L). The variables RM and L were the most significant in the process. An increase in alcohol concentration from 1:2 to 4:1 caused a decrease in yield to 18.45%, after 48 h of reaction time, and a variation of L from 1 to 10% caused an increase in yield to 32.2%. The optimum conditions found to achieve maximum ester yield (75%) are: RM = 1.5:1; L = 5.5% (w/w), 40 degreesC, after 48 h of reaction time. (C) 2003 Elsevier Ltd. All rights reserved.