The phytotoxic activity of cafenstrole [1-(diethylcarbamoyl)-3-(2, 4, 6-trimethylphenyl sulfonyl)-1, 2, 4-triazole] on rice and barnyardgrass (Echinochloa crus-galli (L.) Beauv, var. formosensis Ohwi) seedlings emerging from various soil depths was investigated in relation to its behavior in the soil. when cafenstrole was applied into flooding water under water leakage conditions, the growth of rice was markedly inhibited when the seedlings emerged from the top soil layer but scarcely inhibited when they emerged from deeper soil layers. In contrast, the growth of Barnyardgrass was remarkably inhibited. In barnyardgrass, the shoot basal part and coronal roots grew in association with the first internode elongation in the soil surface layer from seeds located in layers of various depths. Herbicide-treated layer was clearly formed in the top layer and the herbicide present in the soil water was found only in this layer when cafenstrole was applied into flooding water under water leakage conditions. After soil-mix application with cafenstrole, the herbicide was distributed almost homogeneously in all soil layers, and a sufficient concentration to inhibit plant growth was retained in the soil water of each layer. In this system, the growth of rice seedlings emerging from each layer of soil was similarly inhibited, and not dependent on the emergence depth, as true of barnyardgrass. It was suggested that the phytotoxic activity of cafenstrole on the growth of barnyardgrass depended on its absorption from the herbicide-treated layer through the basal part of the shoot, coronal roots and internode with adventitious roots, regardless of the emergence depth. In addition, the low phytotoxic activity on the growth of rice when the plants were transplanted below the herbicide-treated layer was attributed to the absence of the absorptive parts in the herbicide-treated layer.