Plackett-Burman screening, factorial designs, and second-order response surface methodology. (RSM) for medium optimization were employed for lovastatin production by a high-producing mutant of Aspergillus terreus. The method used was effective in screening for nutritional requirements in a limited number of experiments. In a 7-day fermentation, a 4.18-fold increase (i.e. 1.05 g/l) was obtained at (g/l): lactose (89.5), glycerol (30.2), peptone (6.0), soybean meal (SM) (4.2), glucose (20.0), and corn steep liquor (CSL) (10.0), respectively. In the procedure of medium optimization, the formation of compact and highly shaved-off pellets of A. terreus was critical for lovastatin production, where the optimal average core size was about 0.85-1.05 mm. The lovastatin-producing phase was further prolonged if feeding 0.1 g/l D- or D,L-methionine at 72 h. This resulted in a 20% increase at day 10. Feeding provided no benefit if this dosage was added initially, and the addition also did not stimulate lovastatin production while added at late idiophase. Special attention was given to an adverse effect if using L-methionine. (C) 2003 Elsevier Science Ltd. All rights reserved.