Iterative dynamic programming employing randomly chosen candidates for admissible control is applied to the minimization of distillation startup time. The solution of this high-dimensional problem was facilitated mainly by avoiding iterative algebraic equation solving inside a rigorous dynamic distillation model, thereby reducing the computation time. Several illustrative applications of various complexity degrees are presented. The control variables are the reflux ratio, the reboiler heat duty, or both. Both piecewise constant control and piecewise linear control strategies were employed. The decrease of the startup times and heat consumption is high, having an appreciable economic significance.