Previous studies in the Japanese mountains have suggested that slope failure plays a significant role in determining the structure of drainage networks. However, few studies have examined the ef-fects of slope failure on Horton's parameters of drainage nets such as the bifurcation ratio and the stream length ratio. To address the question, this paper examines the drainage nets of mountain river basins in the Matsumoto region of central Japan, which are subject to frequent slope failure (Fig. 1). Two types of drainage nets were constructed on topographic maps: “D-type drainage”-iden-tified from deeply notched v-shaped contours with angle<53° and “T-type drainage”-flow lines along shallower valleys combined with the D-type drainage (Figs. 2 and 3). The comparison between the constructed drainage nets and geomorphological maps has revealed that the D-type drainage mostly occurs in deep valleys cutting into the bedrock, whereas the T-type one generally occurs both in the deep valleys and in shallow hollows caused by regolith failure (Table 2).<br> Horton's law of stream numbers and lengths generally holds for both D-type and T-type drainages (Figs. 4 and 5) . Mean bifurcation ratios for both drainage types exceed the theoretical values report-ed in previous works (4.0), and occasionally reach 5.0. Mean stream length ratios tend to be less than 2. 0, although the theoretical ratios are 2.0 to 2.3 (Table 3) . The T-type drainage has larger bifur-cation ratios and smaller stream length ratios than the D-type one, showing stronger disagreement with the theoretical values.<br> Further analyses on both types of drainage nets have shown that “excess streams”, which do not con-tribute to increase in stream orders despite their joining, abundantly exist on the valley-side slopes of high-order streams (Table 4). The number of excess streams is larger than that of the most proba-ble drainage nets deduced from mathematical theory. This fact accounts for the increased number of low-order streams, and thus for the large values of mean bifurcation ratios. Stream length ratios be-tween 1st- and 2nd-order streams are much smaller than those between higher-order streams (Fig. 5, Table 3), indicating that the low values of mean stream length ratios reflect the notably large length of 1st-order streams. These tendencies are more distinct for the T-type drainage than for the D-type one.<br> The unique drainage structure in the study area is ascribed to the frequent occurrence of storm-induced slope failure, the most dominant erosion process in the region. Because slope failures left many hollows and gullies on valley sides of high-order streams, the number of low-order streams in-creased. Failures also led to the migration of channel heads toward upper hillslopes, resulting in elon-gated 1st-order streams. Because the T-type drainage reflects the effects of shallow failure more direct-ly, it has larger bifurcation ratios and smaller stream length ratios than the D-type one.<br> To examine the validity of the above inference, the known values of the bifurcation ratio and the stream length ratio for various regions were compiled (Table 6). Most of the values had been calculat-ed for the drainage nets including small channels, identified by T-type contour crenulation, field sur-veys, and aerial-photo interpretation. The compilation has revealed that the bifurcation ratios and the stream length ratios for the Matsumoto region are comparable to those for other Japanese moun-tains. By contrast, river basins subjected to less frequent slope failure, such as those in Japanese non-mountainous areas and foreign countries, have smaller bifurcation ratios and larger stream length ratios than those in Japanese mountains. These findings validate the hypothesis that widespread slope failure leads to larger bifurcation ratios and smaller stream length ratios.