Fecal steroid hormone of 36 male captive forest musk deers (Moschus Berezovskii) in Maerkang Musk Deer Farm (MMDF), Sichuan Province was measured based on non-invasive sampling and radio immunoassay (RIA) methods from September 1^st to October 15^th, 2018. In combination with data of musk production, the relationships between fecal testosterone, cortisol and musk secretion were analyzed. The results showed that, musk secretion (12.53±0.76) g (n=36) was significantly positively correlated with fecal testosterone (106.03±12.34) ng/g (n=36) in MMDF's captive forest musk deer. The testosterone concentration (141.51±21.86) ng/g (n=17) of male musk deer with over-averaged musk (16.50±0.61) g (n=17) was significantly higher than that (74.27±7.93) ng/g (n=19) of with below-averaged musk (8.98±0.59) g (n=19) (t=-3.252, df=34, P < 0.01), and the equation y=0.021x+10.270 (R²=0.120, P < 0.05) could fit the relationship well. No significant correlation between musk secretion and the cortisol concentration (47.49±1.53) ng/g (n=36) (r=0.078, P > 0.05), and cortisol concentration (48.98±2.28) ng/g (n=17) of male musk deer with over-averaged musk was slightly higher than that of with below-averaged musk (46.17±2.14) ng/g (n=19), but not significant (P > 0.05). The equation y=0.038x+10.730 (R²=0.023, P > 0.05) approximately fitted the relationship between cortisol concentration and musk secretion. In addition, testosterone concentration significantly affected the musk secretion (P < 0.05), and musk secretion (14.97±1.17) g (n=12) with higher testosterone concentration was significantly more than that (11.32±0.90) g (n=24) of with lower testosterone level (F=4.79, P < 0.05), while the cohort effects of testosterone, cortisol and age on musk secretion were not significant, neither were the interactions. The definite relationships between testosterone, cortisol and musk secretion can be used in musk production prediction and strain selection.