Abstract The changing market forces are driving olefin producers to operate their steam crackers at high operating rates and optimum production performances. These require the use of an optimizer for assessment of viable operational strategies. Most of the optimizers currently employed in this area make use of the coil outlet temperature to determine an improved severity. In rare cases only, a multivariable optimization is performed on the furnace operational conditions. This kind of optimization is implemented in PyroFur for optimizing both the furnace operation and the reaction severity using a set of first-principles mathematical models in which the coil outlet temperature, the total feed flow rate and the steam/hydrocarbon ratio are altered. The suggested procedure was applied as a case study for a typical liquid furnace. Typical improvements obtained from the optimizing control strategy proposed in this article include about 1.4% increased profitability per year per furnace. Effect of a change in the prices or the feed composition is also presented.