To document the impacts of windthrow in riparian leave strips and identify the components needed for small stream large woody debris (LWD) recruitment modeling, we monitored nine small streams at a temperate rainforest site in coastal British Columbia. This study was a component of a larger integrated study of forest management impacts on small streams. A series of small clearcuts were harvested in 1998 in a 70-year-old second growth stand that had regenerated naturally following logging and wildfire. Three cutblocks each were assigned to 10 m and 30 m buffer width treatments and three areas were assigned as unharvested controls. Seven years after the 1998 logging, all logs greater than 10 cm diameter that spanned at least part of stream channel width were measured. A total of 179 logs were recorded. Post-harvest windthrow was higher in the 10 m buffer treatment, while competition-related standing tree mortality was higher in the controls. The major windthrow events had occurred in the first and second years after logging of adjacent stands. There was no significant difference in the number of spanning and in-stream logs in the 10 m, 30 m buffer and control treatments. More than 90% of the LWD was in the 10–30 cm diameter classes. The majority of logs were oriented perpendicular to the stream channel. At the time of measurement, the majority of these trees were still suspended above the stream channel, indicating that the recruitment of logs into the stream channel is a long-term process. Time to recruitment into the channel is dependent on log and valley geometry, log size, species, and log condition prior to toppling. Log height above stream was negatively correlated with log decay class and valley width. Log length was negatively correlated with state of decay, and many windthrown logs were in an advanced state of decay before they entered the stream.