HP turbine usually works in complex operation conditions and subjects to strong temperature variability and thermal gradients. For these reasons, some small variations caused by manufacturing process and operation conditions will yield great effect on the heat transfer performance and the life of hot components. UQ is a state of the art that seeks to provide a framework for calculating the effects of input parameter uncertainty on model output variables. This study presents an investigation of the uncertainty quantification (UQ) of conjugate heat transfer of a typical high load HP cooled turbine vane, considering the influence of total inlet temperature, turbulence intensity and different surface roughness level. The non-intrusive Probabilistic Collocation Method, integrated with sparse grid quadrature has been used in the UQ analysis. The scaled sensitivity analysis is also carried out to assess the influence of each input uncertainty on the output response. The UQ analysis implies that the increasing of surface roughness will enlarge uncertainty of metal wall temperature. As a result, it needs to amplify the safety margin coefficient in order to assess the turbine guide vane surface temperature distribution and service life.