An airbag upon collision of automobiles must finish expanding at the time when the driver's body arrives at the surface of the expanded airbag from the normal position and it must receive the body softly. Either early or late expansion reduce the effect. Determination of the optimal timing for trigger the airbag is quite important. But it is a difficult job, due to the time delay in the bag filling with the gas after collision.This paper focuses on the issue of the airbag technology to how to find the optimal trigger timing and presents a new and straightforward algorithm to determine the timing by introducing the concept of the prediction. The algorithm first predicts how the driver's body will move in the future time for a time equivalent to the delay time and triggers the airbag by the predicted information so that it can compensate the delay by the airbag operation.The algorithm determines the timing only from the acceleration measured by one sensor set in the automobile and is composed of the following four basic processing blocks; Kalman filtering to estimate the acceleration of the driver's body and its time derivative from the noisy measured acceleration. Filtering to estimate the velocity and displacement of motion of the driver's body. Prediction of the velocity and displacement of motion of the driver's body after collision. Judgement to trigger the airbag from the above variables estimated and predicted variables. Simulations in the various conditions were carried out and the results demonstrated the validity of the algorithm.