Abstract: This short paper presents a simple technique to increase the efficiency of random walk algorithms for radiosity. Each generated random walk is used to simultaneously sample two independent estimators. The first estimator is the commonly used shooting estimator, yielding the radiosity at each destination of a random walk due to selfemitted light at the origin. The second estimator estimates the radiosity gathered at the origin due to selfemitted light at each destination. Closed formulae for the variance of the involved estimators allow to combine the resulting radiosity estimates in a near optimal way. Empirical results are in good concordance with the heuristical prediction, and show an increase in efficiency of up to 49%.