Accurate aircraft landing time predictions provide situational awareness for air traffic controllers, enable decision support algorithms and gate management planning. This paper presents a new approach for estimation of landing times using a tree-based ensemble method, namely Quantile Regression Forests. This method is suitable for real-time applications, provides robust and accurate predictions of landing times, and yields prediction intervals for individual flights, which provide a natural way of quantifying uncertainty. The approach was tested for arrivals at Dallas/Fort Worth International Airport over a range of days with a variety of operational conditions.