To observe the effect of herbal-cake-partitioned moxibustion (HCPM) of "Shenque" (CV8) and "Daheng" (SP15) on abdominal pain, plasma β-endorphin (β-EP), uterine prostaglandin E2 (PGE2) and prostaglandin F2α (PGF2α) levels, as well as splenetic natural killer cell (NK cell) activity in primary dysmenorrhea (PD) rats, so as to explore the specificity of acupoint function and the underlying mechanisms of moxibustion in relieving dysmenorrhea. A total of 40 female rats were randomized into blank control, model, CV8-direct moxibustion, CV8-HCPM and SP15-HCPM groups (n=8 rats in each). The PD model was established by subcutaneous injection of estradiol benzoate injection (0.2-0.5 mg/rat) for 10 consecutive days and intraperitoneal injection of oxytocin (2 U) 24 h after the last subcutaneous injection. Moxibustion or herbal-cake (composed of Radix Angelicae Sinensis, Rhizoma Chuanxiong, Radix Paeoniae Rubra, Cortex Cinnamomi, etc.)-partitioned moxibustion was applied to CV8, SP15 or umbilicus respectively for 7 moxa-cones every time, once daily for 10 successive days. The rats of the control and model groups were also restrained as those in the moxibustion groups. The writhing times within 30 minutes was recorded and the contents of plasma β-EP, uterine PGE2 and PGF2α were detected by ELISA, and NK cell activity was detected using MTT. Compared with the control group, the writhing times and the content of PGF2α in the uterus tissue were significantly increased in the model group (P<0.01), while the contents of plasma β-EP, uterine PGE2 and splenetic NK cell activity were significantly decreased (P<0.01). In comparison with the model group, the writhing times and uterine PGF2α content were obviously down-regulated in the SP15-HCPM, CV8-direct moxibustion and CV8-HCPM groups (P<0.05, P<0.01), and the contents of plasma β-EP and uterine PGE2, and splenetic NK cell activity were significantly increased (P<0.05, P<0.01). The therapeutic effects of CV8-HCPM group were significantly superior to those of SP15-HCPM and CV8-direct moxibustion groups in lowering writhing times and PGF2α level, and in up-regulating β-EP, PGE2 (P<0.05, P<0.01). The NK cell activity of CV8-HCPM group was significantly increased compared with that in the SP15-HCPM group(P<0.05). No significant differences were found between the SP15-HCPM and CV8-direct moxibustion groups in the levels of writhing times, plasma β-EP and uterine PGE2, PGF2α contents and splenetic NK cell activity (P>0.05). Moxibustion of both CV8 and SP15 can relieve abdominal pain in PD rats, which may be closely associated with its effect in suppressing PD-induced decrease of plasma β-EP and uterine PGE2 levels and splenetic NK cell activity and increase of uterine PGF2α. The therapeutic effect of CV8-HCPM is obviously better than that of SP15-HCPM and CV8-direct moxibustion.