In this paper, nonlinear state feedback laws are developed for on-line optimization of batch/semi-batch processes with multiple manipulated inputs. The first-order necessary conditions are cast in terms of the system Lie brackets. Optimal state feedback laws are derived which are independent of the adjoint states. It is shown that the feedback laws are, in general, dynamic in nature and provide an alternative to solving a two point boundary value problem on-line. This methodology is illustrated via simulation of several end-point optimization problems in batch chemical reactors. (C) 1998 Elsevier Science Ltd. All rights reserved.