The abundances of elements in the Earth and the terrestrial planets provide the initial conditions for life and clues as to the history and formation of the Solar System. We follow the pioneering work of Bond et al. (Bond, J.C., Lauretta, D.S., O'Brien, D.P. . Icarus 205, 321-337) and combine circumstellar disk models, chemical equilibrium calculations and dynamical simulations of planet formation to study the bulk composition of rocky planets. We use condensation sequence calculations to estimate the initial abundance of solids in the circumstellar disk with properties determined from time dependent theoretical models. We combine this with dynamical simulations of planetesimal growth that trace the solids during the planet formation process and which include the effects of gravitational and hydrodynamical mixing. We calculate the elemental abundances in the resulting rocky planets and explore how these vary with the choice of disk model and the initial conditions within the solar nebula.Although certain characteristics of the terrestrial planets in the Solar System could be reproduced, none of our models could reproduce the abundance properties of all the planets. We found that the choice of the initial planetesimal disk mass and of the disk model has a significant effect on composition gradients. Disk models that give higher pressure and temperature result in larger variations in the bulk chemical compositions of the resulting planets due to inhomogeneities in the element abundance profiles and due to the different source regions of the planets in the dynamical simulations. We observed a trend that massive planets and planets with relatively small semi-major axes are more sensitive to these variations than smaller planets at larger radial distance. Only these large variations in the simulated chemical abundances can potentially explain the diverse bulk composition of the Solar System planets, whereas Mercury's bulk composition cannot be reproduced in our approach.