The geometry of tetracyanoplatinate(II) (TCP) has been optimized with density functional theory (DFT) calculations in order to compare different computational strategies. Two approximate scalar relativistic methods, i.e. the scalar zeroth-order regular approximation (ZORA) and non-relativistic calculations with relativistic effective core potentials (ECPs), were benchmarked against the four-component fully relativistic approach using the Dirac-Coulomb Hamiltonian and all-electron non-relativistic calculations. We find that the 5% contraction of the platinum-carbon bond due to relativistic effects is almost quantitatively reproduced in the ZORA and ECP calculations. In addition, the effect of the exchange-correlation functional and one-electron basis set was studied by employing the two generalized gradient approximation (GGA) functionals, BLYP and PBE, as well as their hybrid version B3LYP and PBE0 in combination with both correlation consistent and Ahlrichs type basis sets. The platinum-carbon bond length (relativistic or non-relativistic) is approximately 1% shorter on using the PBE exchange-correlation functional compared to the BLYP functional but including exact exchange has no significant effect. For the C-N bond these trends are reversed and an order of magnitude smaller. With respect to the basis set dependence we observed that a triple zeta basis set with polarization functions gives in general sufficiently converged results, but while for the Pt-C bond it is advantageous to include extra diffuse functions, this did not turn out to be important for the C-N bond. Keywords: 4-Component calculations, DFT, geometry optimization, relativistic effects, transition metal complexes, ZORA.