As the number of World Heritage Sites (WHSs) is time-invariant or rarely changing in the short run, the panel data model reaches inconsistent estimates of the real effect of WHSs on tourist arrivals. This study adopts two alternative techniques, the fixed-effect vector decomposition (FEVD) approach and the two-stage, double fixedeffect (TSDFE) model, to revisit the WHS-Tourism nexus. Based on China's provincial-level panel data over the 2000-2005 period, estimates using the fixed-effect panel data model show an insignificant impact of WHSs on tourist arrivals. Crucially, the empirical results obtained from both FEVD and TSDFE confirm a positive relationship between WHSs and international tourists in China. Moreover, cultural WHSs tend to exhibit a stronger tourism-enhancing effect than the natural WHSs. This study also highlights the importance of dealing with the time-invariant and/or rarely changing attributes in examining tourism demand.