This paper presents an optimization-based dynamic simulation for pregnant women pseudo standing, forward falling, and pulling tasks. Based on anatomy of pregnant women a digital pregnant woman model is developed. The model has 55 degrees of freedom (DOFs) including 6 global DOFs and 49 body DOFs. Recursive dynamic algorithm is used to formulate the equations of motion. Human motion can be formulated as a non-linear optimization problem. Control points of B-spline curves that represent joint angle profiles are design variables. The joint angles, angular velocities and angular accelerations, will be obtained from the control points. The summation of all joint actuator torque square acts as the objective function. Besides some common constraints, different constraints are adopted for standing, falling, and pulling, respectively. Three cases, non-pregnancy, 6-month, and 9-month pregnancy, are investigated. For the pulling task, 2N, 100N or 200N external load is applied as the pulling force. Determinant joints (hip, knee and ankle) are plotted to analyze the simulation results. The simulation results show the effects of pregnancy on human movement kinematics and dynamics. The average computational time for each case is close to 3.5 minutes in a Dell computer with 3.25 GB of RAM and 3.16 GHz.