The present study proposes two correlations for the prediction of heat transfer to supercritical water for the cases with given heat flux and given wall surface temperature, respectively. Three steps are taken to develop the new correlations. At first a large database was established. The reliability of each test point in the database was assessed with respect to its consistence and reproducibility. In the second step, important dimensionless parameters were identified with two different approaches, the Spearman's rank correlation and the group-wise statistic assessment. Both approaches led to the similar outcomes and identified the most important dimensionless parameters, which can be used to predict the heat transfer coefficient. In the third step, two correlations were proposed for the cases with given heat flux and given wall surface temperature, respectively, to avoid iterative procedure and subsequently to eliminate possible multiple solutions. Comparison of the new correlations with an extensive database consisting of test data in various supercritical fluids such as water, R134a and CO2 confirmed the applicability of the new correlations in a wide range of parameters. In addition, the results indicate that in general the correlation for the cases with given wall surface temperature shows much better prediction accuracy than that for the cases with given heat flux.