Tolerance of Pichia anomala, a strain of yeast associated with olive fermentation, to salt, temperature, and pH was studied in yeast-malt-peptone-glucose medium using a nonfactorial central composite experimental design with three repetitions in the center to account for pure error. Modified Gompertz, logistic, Richards-Stannard, and Baranyi-Roberts models were used to determine maximum specific growth rate (micro(max)) and lag phase period (lambda) from the growth curves (primary models). All models produced a good fit (significant at P < 0.05), but the graphical and statistical analyses of the data indicated that the modified Gompertz and Richards-Stannard models were the most appropriate. The biological parameters obtained with the diverse models were fitted to a response surface secondary model. A significant decrease in micro(max) was observed as temperature decreased and salt increased. A significant increase in lambda was observed as temperature (linear and quadratic effects) and pH decreased and as salt content increased. Effects of interactions were complex and depended on models. Validation revealed acceptable errors and bias in micro(max) and lambda values obtained in independent experiments. Validation growth curves were best reproduced by using the values of micro(max) and lambda predicted by the response surface from the modified Gompertz and Richards-Stannard models. Results from this study can be applied to table olive fermentation or storage and for production of table olives as refrigerated commercial products without the use of preservatives or pasteurization.