A systematic design procedure was developed to aid in the reduction of trial and error in the design of reactor control systems. The method developed here uses an automatic optimization to help the designer find the best controller structure and the best controller settings to use in a control system based on conventional controllers (derivative, proportional, or integral). The method was demonstrated in a control system design for a molten salt breeder reactor. The application of the method gave excellent results at only a modest expense in computer time (5 min on the IBM-360 Model 91). The method appears suitable for routine application in reactor control system design.