A culture medium for increased xylanase formation by a wild strain of Sclerotium rolfsii was optimized by applying statistically designed experiments. The optimization process was divided into three basic steps. In the initial phase of screening, two different fractional factorial plans, a Graeco-Latin square design and a folded Plackett-Burman design, were employed. From the list of medium components the relevant variables for xylanase formation (cellulose. peptone, and trace elements) were identified. The second step of optimization used a central composite experimental design to calculate a predictive model. In this phase only the two most important factors, i. e. cellulose and peptone from meat, were considered. The third step of verification validated the results of the optimization. Optimal con centrations of cellulose and peptone were found to be 42.6 and 80.0 g l(-1), respectively. Shaken flask cultivations of S. rolfsii using the optimized medium yielded a maximum xylanase activity of 394 IU ml(-1) (6,570 nkat ml(-1)) within 13 days. Growth on the optimized medium also resulted in elevated levels of other hemi-cellulolytic enzyme activities including mannanase, alpha-arabinosidase. and acetyl esterase. Especially the value of 155 IU.ml(-1) (2,580 nkat ml(-1)) for mannanase is remarkable since it appears that this is among the highest activities reported for fungal organisms.