The Plackett-Burman design and the response surface method (RSM) with a central composite design (CCD) were employed to develop a culture medium for ansamitocin P-3 production using Actinosynnema pretiosum ATCC 31565. Among the 11 nutrients tested using the Plackett-Burman design, two carbon sources, sucrose and dextrin, and two nitrogen sources, polypeptone and yeast extract, were selected. Optimization of the concentrations of the selected nutrients was then performed using RSM with CCD. After two rounds of RSM, the optimum concentrations ($\%w/v$) of sucrose, dextrin, polypeptone, and yeast extract were identified as 4.5, 4.5, 0.16, and 0.89, respectively. The maximum ansamitocin P-3 titer was 45.2 mg/l with the optimized medium, which was about 6 times higher than that (7.315 mg/l) obtained with an $R_{2}YE$ medium before optimization.