The R-squared for multiple regression is positively biased. Bootstrap can be utilized to simulate the empirical distribution of the R2 and remove its estimation bias. The bootstrapping method does not require strong distributional assumptions. An example of multiple regression is included to visualize the estimation bias in the R2. Compared to the bias estimate from the adjusted R-squared, the bias estimate based on bootstrap is smaller. So, the bootstrap bias corrected R2 is rightfully between the regular R2 and the adjusted R-squared.