To optimize ginsenosides hydrolyzing beta-glucosidase production from Aspergillus niger, response surface methodology was carried out in two stages. The Plackett-Burman design was achieved to screen the important variables that influence beta-glucosidase production. Among 10 variables (wheat bran, soybean powder, CaCl(2), ginsenosides, KH(2)PO(4), MgSO(4), polyethylene glycol (PEG), medium volume, inoculum size, and stirring speed), it was found that wheat bran, KH(2)PO(4), and stirring speed had significant effect on beta-glucosidase activity due to very low p-values (p<0.05). Subsequently, wheat bran, KH(2)PO(4), and stirring speed were further optimized using central composite design. The optimal beta-glucosidase production was predicted to be 4650.14 U/ml with the combination of factors (wheat bran, 34.51 g/l; KH(2)PO(4), 1.78 g/l; stirring speed, 161.60 rpm/min). Finally, under optimal fermentation conditions, ginsenoside Rb(1) was converted to Rd and F(2) by A. niger within 10 min. Little compound K was detected at 30 min, and finally F(2) was completely transformed to compound K within 8 h. The putative conversion pathway of Rb(1) by A. niger was Rb(1), Rd, F(2), and compound K.