The rapid adoption of the basic oxygen steelmaking process by American steelmakers has been accompanied by an increasing desire to improve control of the process. Digital computing equipment is the focal point of most efforts toward improving process control. Because of the notable lack of process feedback instrumentation, development of accurate mathematical process models is vital to the success of computing control applications. A dynamic mathematical model of the removal of carbon from the steel batch is developed, based on present knowledge of process rate limiting steps. The resulting nonlinear model contains several unknown constants; the pattern search optimization technique is employed to establish values of these constants from production data. The fitted model comprehends the variation in carbon removal of an actual oxygen steelmaking process.