Estimation of aircraft weight in flight, which is a dominant parameter related to its flight performance, was studied. A typical approach for the estimation is to use a simple combination of initial weight on the ground and fuel consumption in air obtained with fuel tank gauge or by accumulating fuel flow. This approach is insufficient when the flight performance will be estimated as accurately as possible because some of the available measurements are not utilized. Therefore, the forward–backward smoother derived from Kalman filter was applied to the estimation with our revision of the smoother to fuse terminal weight on the ground additionally. According to numerical simulations, our method estimated the weight in smaller errors than not only the typical approach but also a fixed-interval smoother, which is used generally for an off-line estimation problem. Moreover, application to actual flight data showed that our method improved the estimated standard deviation by approximately three percent at maximum compared to the fixed-interval smoother.