Chemical and mechanical properties of the welded joint are greatly affected by flux composition. The resulting chemical compositions of C, Si, Mn, P and S in the weld metal have been studied using formulated fluxes. The constrained mixture design, extreme vertices design, has been used to formulate fluxes to study the effect of flux constituents. Regression models were developed for weld metal content in terms of individual flux constituents and their binary mixtures for submerged arc welding of high-strength low-alloy steel. From the results, it is found that CaF2 is the most significant flux constituent and Al2O3 is the second most significant constituent among individual mixtures. CaO-MgO and CaO-Al2O3 binary mixtures are the most effective to change weld metal content. Regression mathematical models have been checked for adequacy using t-test and analysis of variance (F-test). Flux mixtures' composition has been provided for optimum chemical composition of weld metal. © IMechE 2014.