Sequential statistical experimental designs (Plackett-Burman, factorial, response surface and ridge analysis) were used to improve the medium for the production of Echinocandin B (ECB) by Aspergillus nidulans. Special attention is given to the ridge analysis, a search and optimization technique which can be applied to second-order models. Fifteen medium ingredients were investigated using a total of 108 shake flask fermentations over five stages of experimentation. The iterative process of thoughtful design and careful data analysis ultimately produced a simple, five component media, which increased titre by 46%. (C) 2001 Elsevier Science Ltd. All rights reserved.