The combined effects of initial xylitol supersaturation value (Xyt(s)) and cooling temperature (T-c) on xylitol crystallization from synthetic solutions were investigated using response surface methodology. A 3(2) full-factorial design was employed for experimental design. Synthetic solutions were subjected to crystallization, aimed at contributing to the development of a complete biotechnological process, including preliminary xylose-to-xylitol bioconversion in hemicellulose hydrolyzate and final xylitol recovery by crystallization. The crystallization methodology consisted of concentration of synthetic solutions up to supersaturation, cooling of the supersaturated solutions, separation of crystals by centrifugation, and final filtration. By means of response surface analysis, the statistical model identified the operating conditions (Xyt(s) = 728 g/l and T-c = -6.0 degreesC) under which purity degree (0.97) and xylitol crystallization yield (0.54) were simultaneously optimized. On the other hand, a purity degree close to 1.0 to meet industrial standards is expected at Xyt(s)= 583 g/l and T-c = -2.4 degreesC, but the crystallization yield would be unsatisfactory (0.25). (C) 2003 Elsevier Ltd. All rights reserved.