Scaling-up purine nucleoside fermentation by a mutant strain of Bacillus subtilis from a shaking flask to a stirred-tank fermentor was attempted. The dimensions and the operating conditions of the stirred tank were determined in order to satisfy the optimum conditions of O2 transfer and power consumption per unit volume for the shaking flask. When the purine nucleoside fermentation was carried out in the stirred-tank fermentor under these conditions, in which the temperature simulated that in the shaking flask, the total amount of purine nucleosides produced was almost the same as that in the shaking flask, but the accumulation ratio of guanosine to total nucleosides was different from that in the flask. Since urea could not be utilized so efficiently in the stirred-tank fermentor, the NH4+ concentration and the pH of the culture broth were lower than those in the shaking-flask culture during fermentation. The activity of inosine monophosphate dehydrogenase and the accumulation ratio were significantly affected by the NH+4 concentration. When the pH of the stirred-tank culture was maintained at 6.9 by ammonia water to keep the NH+4 level higher, the ratio was improved to the same level as that observed in the shaking-flask culture. The fermentation heat calculated from the shaking-flask data and its pattern of change were similar to those in the stirred-tank fermentor.