The dynamic optimization of batch crystallizers has been widely studied. A simplified method of optimization inspired from nonlinear model predictive control with a receding horizon is introduced and tested on many different objective functions with various constraints. The proposed optimization method with a receding horizon gives excellent results with no noticeable difference from those obtained by rigorous dynamic optimization and is well adapted to online dynamic optimization. Two different crystallizer models are compared. It is shown that the dynamic optimization problem is constituted of several subproblems related to the constraints on the crystallizer temperature, on the concentration compared to the metastable concentration or on the final moments. Finally, the authors propose simple online control algorithms that result in quasi-optimal temperature profiles provided that the type and sequence of arcs have been previously determined. This method is well adapted to industrial situations. (C) 2008 American Institute of Chemical Engineers AIChE J, 54: 3188-3206, 2008