The speed and precision of alignment can affect the navigation precision and reaction speed of weapon launch. Aimed at missile carried in aircraft, the research designs inertial sensor models and rapid transfer alignment considering body flexure and data delay on dynamic base. In the process of sensor modeling, this study firstly build angular rate and acceleration models according to the aircraft maneuver. Then sensors’ errors including physical misalignment, flexure, gyros drift and accelerometer bias are added to these models, as the outputs of sensors using in weapon INS (inertial navigation system) and aircraft INS. In the design of transfer alignment, the weapon INS firstly updates velocity and attitude by itself, with initial values transferred from aircraft INS. Thanks to the existing of errors, a lower dimension Kalman filter is designed to estimate the attitude errors and sensor errors of alignment to calibrate weapon INS. The body flexure and data delay are considered and solved in the process of alignment. The whole simulation is implemented in MATLAB. The experiment proves that the alignment time is less than 20 seconds, while the alignment precision can arrive at 3 mrad. At the same time, effects of data delay and weapon INS noise are analyzed in the simulation.