This study proposes a method for the estimation of lag time of forested mountainous watersheds. Hydrological flood-flow design methods require a time parameter, such as lag time or time of concentration, to estimate peak discharge and the flood hydrograph shape. Most existing methods used to compute the time parameter of a watershed are empirical and have been developed for urban or rural watersheds. Use of these methods for forested mountainous watersheds may result in severe underestimation of watershed response time and consequently, in overestimation of peak flood discharge. The water flow in a watershed is separated and analysed in two phases,viz.the land or hillslope phase and the stream channel phase. Most of the flow in a forested steep watershed is generated through subsurface pathways, and this knowledge, acquired from field experiments, is combined with kinematic wave equations to describe the generation of flow from steep forested hillslopes. This hillslope runoff is then used as input to the stream channels. Kinematic wave equations are developed for the runoff movement in the channels by assuming that both the roughness coefficient and the stream channel slope vary with the distance from the outlet of the watershed. These assumptions are validated and confirmed with data from the USA and Coastal British Columbia, Canada. The resulting equations are integrated to obtain an equation for the lag time. Comparison of the results of the proposed equation with data from two experimental watersheds in Coastal British Columbia and empirical equations used for the calculation of the lag time indicates that the proposed method is reliable and gives a good approximation of the observed lag time. These results are compared with other empirical equation, and it is shown that these earlier methods can result in severe underestimation of lag time.