Shape control in producing thin cold-rolled strip is very complex because of the non-linearity of the process. Depending on the process conditions, the operator is involved with manual control during production in the steel industry. In order to implement the operator's knowledge in the shape control, a fuzzy controller and neural network emulator have been developed. The fuzzy-control system that has been developed has been utilized for simulations of cross-sectional shape control in the six-high cold-rolling of thin steel strips of less than 0.5 mm in thickness in the present investigation, the simulations being carried out on an IBM PC 486. The fuzzy logic was created based on production data to control the delivery shape at the last stand of a tandem cold mill. Steady- and non-steady state control simulations of irregular cross-sectional strip shapes and the stability of the currently applied fuzzy control scheme have been investigated. The currently applied fuzzy control scheme is found to be successful in reducing the irregularity of the cross-sectional shape of the cold-rolled thin steel strip in a stable manner for the steady and the non-steady state under the present condition. Thus, the fuzzy-control system might be useful in controlling the process as does a human operator, without introducing manual intervention in practice.