This article investigates the impact of initial condition (specifically, the deviation of the initial observation from the underlying deterministic component) on the power properties of the nonlinear Kapetanios- Snell-Shin (KSS) unit root test. While the Generalized Least Squares (GLS)-detrended KSS test has significant power advantage over the Ordinary Least Squares (OLS)-detrended KSS test if the initial deviation is small, the opposite occurs if the initial deviation is large. These results are similar to those found previously for the Dickey-Fuller (DF) test. To maintain good power properties across different initial conditions, a Union of Rejections (UR) strategy is adopted; the strategy is demonstrated to be useful.