Response surface methodology was used to predict the glucosylation yields of thiamin using immobilized beta-glucosidase. A Central Composite Rotatable Design (CCRD) of 32 experiments with immobilized beta-glucosidase, thiamin, incubation period, buffer concentration and pH, as five independent variables was employed at five levels. A second-order polynomial equation was developed, the regression coefficient values of which exhibited a R-2 value of 0.74. Contour plots explained the glucosylation behaviour of the enzyme through a reversal in glucosylation at a cross-over point corresponding to 60% (w/w D-glucose) immobilized beta-glucosidase and 0.12 mM buffer concentration at pH 6. The highest conversion yield of 58% obtained experimentally compared well with the predicted yield of 52% under optimum conditions of 40% immobilized beta-glucosidase, 0.55 mmol thiamin, 96 h incubation period and 0.16 mM buffer concentration at pH 7. Validation experiments carried out at certain random predictive conditions also showed good correspondence between experimental and predictive yields. (C) 2008 Elsevier Ltd. All rights reserved.