The most appropriate time to introduce androgen deprivation therapy for prostate cancer remains controversial. Our aim was to evaluate the effects of early versus delayed surgical castration on prostate cancer progression and survival in the transgenic adenocarcinoma of the mouse prostate (TRAMP) model. TRAMP mice were randomly divided into three groups: the early castration group (on which castration was performed at the age of 4 weeks), the delayed castration group (on which castration was performed when abdominal turnouts could be palpated), and the sham-castrated group. Mice were monitored daily throughout their lives until cancer-related death or the develop-ment of an obviously moribund appearance, at which time the individual mouse was killed. Androgen receptor expression in prostate turnouts was also evaluated. The results shows that the average lifespan in early castration, delayed castration and sham-castrated groups were 54.1 weeks, 59.9 weeks and 39.1 weeks, respectively. Both early castration and delayed castration conferred a statistically significant survival advantage when compared with the sham-castrated group (P<0.001). However, the difference in lifespan between the early castration group and the delayed castration group was not statistically significant (P=0.85). The increase in lifespan in the TRAMP mice that received either early or delayed castration correlated with lower G/B value (genitourinary tract weight/body weight) at death than the sham-castrated mice. In conclusion, early and delayed castrations in TRAMP mice pro-longed survival to a similar extent. This finding may provide a guide for clinical practice in prostate cancer therapy.