Temper rolling is essential for the quality of steel sheets. The degree of temper rolling determines the mechanical properties of the steel sheet and is highly influenced by the rolling force or strip tension. Since mathematical models generate unsatisfactory results for the calculation of these two process parameters, other methods for the presetting of tempers mills must be used. The parameter presetting of temper mills is of prime importance because it reduces the effort of tuning these parameters later. Hence, the production costs are reduced by minimizing the amount of wasted material that does not fulfill the quality requirements. Genetic Programming (GP) is an evolutionary inspired and population based modeling technique and has been successfully applied in different contexts. In this paper we present first results of advanced genetic programming concepts on large datasets from a temper mill in comparison to linear regression (LR), support vector machines (SVMs) and previous analysis on the datasets. The use of GP shows an improvement compared to previous work, but is still inferior to models obtained by SVMs. A major advantage of GP compared to support vector machines is that the identified models are mathematical formulae which can be interpreted and enable knowledge generation about the temper rolling process.