Runoff generated from rainfall in a forest watershed can be characterized into surface runoff, interflow, and baseflow. When there is no rainfall for a long period of time, runoff is mainly composed by the baseflow, in which the hydraulic characteristics of soil and parent material determine the recession. Methods of baseflow recession analysis include graphical method, analytical method and others. Under the assumption of single aquifer of watershed, analytical method expresses the baseflow recession characteristics by either linear or nonlinear equations. In this study, we gathered 48 baseflow-event records between 2000 and 2013 in the Lienhuachih experimental watershed No. 2 administrated by the Lienhuachih Research Center, Taiwan Forestry Research Institute. Each event of records was composed of an initial daily discharge starting from the 3^(rd) day after rainfall ceased to at least 12 to 15 days without any rainfall. Three equations, exponential equation, inverse quadratic equation, and reciprocal equation, were respectively applied to calculate the recession coefficients of each event. Then the obtained recession coefficients of each event were used to calculate discharge Qc(t) on the t day by three equations, respectively. The average relative error (ARE) and root mean square error (RMSE) calculated from observed discharge Q_0(t) and Qc(t) were used to evaluate the performance of these three equations. The results showed that the estimated exponential equation, inverse quadratic equation and reciprocal equation were Q_t = Q_0e^(-0.0635t), Q_t = Q_0(1+0.0392．t)^(-2) and Q_t = Q_0/(1+0.1145．Q_0．t), respectively. Their average ARE of 48 events by estimated exponential equation, inverse quadratic equation, and reciprocal equation were 0.075, 0.054 and 0.043 and average RMSE were 0.025, 0.008 and 0.007, respectively. According to the t-test of two population mean difference, the difference of ARE or RMSE between reciprocal equation and inverse quadratic equation were not significant; however, they were significant between exponential equation and the other two equations (P-value ＜ 0.05). Based on the assumption of inverse quadratic equation that the recession of baseflow is dominated by the gravitational drainage of soil layer and shallow unconfined aquifers, and saturated soil hydraulic conductivity decreases with the increasing of depth, results of this study were consistent with previous research in the study area. As a result, this study suggests using the inverse quadratic equation when analyzing the baseflow recession of Lienhuachih experimental watershed No. 2.