The design of controllers for a continuous selection technique (BOICS; Brown and Oliver, 1982) is considered. This technique is used to obtain microbial mutants that are tolerant to extreme environmental stress. Applications of BOICS have been hampered by the problem of controller design. In this paper, a modified implementation of BOICS is considered which has a number of practical advantages. A model-based approach to controller design is taken. The case in which the stress is due to an inhibitory substance in the growth environment is considered. The analysis is intended to be applicable to any reasonable combination of organism and inhibitor. Conventional linear and time-invariant controllers are considered. Guidelines for the selection of controller parameters＇ values are suggested. The application of these guidelines requires that certain process parameters＇ values be identified. Methods by which these parameters＇ values can be identified are suggested. Simulation results indicate that the resulting controllers perform satisfactorily. This is confirmed by experimental data from a model selection experiment. A recipe for the design of controllers is a necessary part of a protocol for BOICS. It is hoped that the solution to the controller design problem that is offered in this paper will encourage further applications for the technique.