Seven kinds of suppositories were constructed with oleaginous base materials (Witepsol H-15 (H-15) and E-85 (E-85)) : a conventional type suppository containing valproic acid (VPA) mixed with E-85 (I), a conventional type suppository containing sodium salt of VPA (sodium valproate) (S-VPA) mixed with H-15 (II), hollow type suppositories containing VPA in the forms of oily liquid (free acid) (III), macrogol 1000 or 6000 mixture (IV or V), powder (S-VPA) (VI) and aqueous solution (S-VPA was dissolved in 0.9% NaCl solution) (VII) in each cavity. The content of VPA in type I was decreased considerably by volatility and type II was found to be hygroscopic. Therefore conventional type suppositories containing VPA or S-VPA were not of practical use, whereas III and VI prevented volatility of VPA and minimized the hygroscopic property of S-VPA. Plasma concentration of VPA was measured in rabbits after rectal administrations of III, IV, VI and VII. By using VI, the highest values of the mean of the peak plasma VPA concentration (C_max) (49.8±2.6 μg/ml) and the mean of the area under the plasma concentration-time curve (AUC) (90.0±3.7 h·μg/ml) were obtained. The C_max and the AUC estimated after administration of VII were not significantly different from those of VI. The C_max and the AUC were lower with III than with IV, VI or VII but the extent of bioavailability (EBA) of III was about 80%. These data on bioavailability suggested that VPA was efficiently absorbed into the rectum of rabbits after the administration of hollow type suppositories. Thus, it was concluded that hollow type suppositories containing VPA or S-VPA can be produced as practical preparations for use.