In a frontal vehicle crash, for a given crash velocity and given maximum vehicle crush, with a known restraint characteristic, what is the vehicle pulse, subject to lower and upper bound constraints, that produces the lowest peak occupant deceleration? A solution procedure using numerical optimization is proposed. The pulse is discretized in the vehicle crush domain. The optimization search is facilitated by a specially developed algorithm that is a hybrid of the sequential quadratic programming (SQP) algorithm for nonlinear constrained optimization and the genetic algorithm (GA). Optimization examples are shown with linear and nonlinear occupant restraints. Numerical results from the examples indicate that when the number of pulse discretization segments is less than five, the solution method is effective in providing pulse improvements for practical problems. A discussion on the theoretical and practical aspects of optimal pulses is also given, with reference to a theoretical optimal pulse recently published by Wu et al. [1].