Beta-D-glucosidase-catalyzed deglucosidation of phenylpropanoid amides of 5-hydroxytryptamine (PAHAs) glucoside in safflower (Carthamus tinctorius L.) seed extracts, including N-(p-coumaroyl)serotonin glucoside (CSG) and N-feruloylserotonin glucoside (FSG), was optimized by response surface methodology (RSM). The Box-Behken design (BBD) was employed to evaluate the interactive effects of independent variables on the deglucosidation rates of CSG and FSG. The variables involved were pH (5.6-6.2), temperature (45-55 degrees C), and enzyme load (2.0-3.0%, relative to the weight of the total substrate). The substrate concentration was fixed at 3.3 g/L on the basis of factorial experiments. The optimum conditions obtained via RSM at a fixed time of 2 h were as follows: pH, 5.9; temperature, 48 degrees C; and enzyme load, 3.0%. Under these conditions, the actual deglucosidation rates of CSG and FSG were 75.5 and 42.2%, respectively, which agree well with the predicted values (75.3 and 41.9%) by RSM. The final incubation time (10 h) was determined by the time course of the deglucosidation under the above-mentioned optimum conditions, which gave the deglucosidation rates of both CSG and FSG above 90%. Simultaneously, 2-hydroxyarctiin, a typical cathartic beta-glucoside, was also removed by 80.3%.