Subsurface flow within forest hillslopes is not well understood. A soil pit containing macropores was prepared and instrumented for subsurface flow measurements at the Hitachi Ohta Experimental Watershed. The subsurface flow from the soil profile was divided into five groups: the porous (including macropores) organic-rich soil layer; three macropore groups in the mineral soil layer; and the mineral soil matrix excluding visible macropores. The outflow from the organic-rich soil layer coincided with rainfall and depended strongly on rainfall intensity. Such a response is related to the location of the organic-rich layer in the uppermost portion of the soil profile. One macropore in the mineral soil was connected directly to a hydraulically active fracture in the bedrock. Therefore, the flow from this macropore group was high and delayed relative to rainfall even though the group is located in the upper left portion of the mineral soil layer. Most of the hillslope discharge occurred from the soil matrix excluding visible macropores. Preferential flow pathways in the mineral soil, as evidenced by staining tests, contributed to stormflow. The flow from two macropores just above bedrock was generally very low and responded gradually to rainfall inputs. This minor outflow was influenced by preferential flow above bedrock. Four macropores in the upper right portion of the mineral soil layer did not connect to preferential flow pathways. Therefore, outflow from this group was measured only when large rain events occurred. Soil moisture, amounts and intensities of rainfall, and bedrock topography affected heterogeneous outflow from the soil profile. The expansion and extension of macropore networks might contribute to stormflow. Results indicate that hydrometric measurements are needed to support inferences derived from end member mixing analysis and topographic-based hydrologic models related to complex preferential flow pathways in forest catchments. Copyright © 2001 John Wiley & Sons, Ltd.