Land surface evapotranspiration （ET_a） is a key element that determines the terrestrial water storage of the Asian Water Tower. However， the long-term variations in ET_a and its response to the ongoing climate change remain largely unknown. Here， this study used both in-situ observation and complementary relationship-based （CR） modeling technique to investigate the changes in ET_a from typical alpine steppe and alpine wetland on the Tibetan Plateau during the last four decades. The results showed that the CR model was able to accurately simulate ET_a once its parameters could be locally calibrated， suggesting that this model has a great potential for understanding the long-term variations in ET_a over such a sparsely-instrumented but hydrologically-important region. During 1973-2013， both alpine steppe and alpine wetland showed increasing trends in ET_a， but such an increase was only significant for the alpine wetland in which ET_a increased with a rate of 2.0 mm/a. Further correlation analysis suggested that the changes in ET_a over these two ecosystems was primarily controlled by changes in the vapor pressure over the last 40 years. The ET_a consistently increased in both alpine ecosystems before the late 1990s， but their changes in ET_a became contrasting after the late 1990s because ET_a decreased significantly over the alpine steppe but continued to increase over the alpine wetland until the mid-2000s. The main reason for the increase in ET_a at the latter ecosystem was the increase in vapor pressure and sunshine hour during this period. Moreover， the soil moisture of the wetland could be replenished from the glacier melting， which could provide enough water for surface evapotranspiration process. This study shows that while the geographical distance is short， the response of ET_a to climate change in these two alpine ecosystems might differ obviously because of the different hydrological cycle regimes.