A response-surface methodology was used to study the effect of carbon:nitrogen ratio (C:N) and initial concentration of total solids (CTS) on insecticidal crystal protein production and final spore count. Bacillus thuringiensis var. kurstaki HD-73 was grown in a stirred-tank reactor using soybean meal, glucose, yeast extract, corn steep solids and mineral sails. Soybean meal and glucose were added according to a central composite experimental design to test C:N ratios ranging from 3:1 to 11:1 and C-TS levels from 60 g/l to 150 g/l. Cry production was quantified using sodium dodecyl sulfate/polyacrylamide gel electrophoresis. The response-surface model, adjusted to the data, indicated that media with a C:N of 7:1 yielded the highest relative Cry production at each CTS The spore count was higher at low C:N ratio (4:1) and high CTS (near 150 g/l). Specific Cry production varied from 0.6 to 2.2 g Cry/10(10) spores. A 2.5-fold increase in C-rs resulted in a six-fold increase of protoxin production at a 7:1 C:N ratio. It is concluded that the best production conditions for Cry and for spores are different and optimization of B. thuringiensis processes should not be done on a spore-count basis but on the amount of Cry synthesized.