In methods of predicting travel time on the Metropolitan Expressway in Tokyo, various reports of research work have been performed so far. Although a so-called pattern matching method is a typical method to predict traveling time, it has a weak point that is not able to predict traveling time with some accuracy under the congestion caused by car accident, rainfall so on. Hence, the authors proposed a method of travel time prediction by Bayesian statistics, which reflected traffic conditions as well as some impacts of unexpected events. However, the cases that traveling time predicted by Bayesian statistics was lower than that of actual were prominent still. It is required to reconsider how to implicate Bayesian statistics. The study proposes that estimated value is improved more accurately by reconsidering how to reform a prior probability with adding factors bringing an expansion of the traveling time.