The unit cell edge length, a, of a set of complex cubic perovskites having the general formula A 2+ 2 BB ′ O 6 is predicted using two methodologies: multiple linear regression and artificial neural networks. The unit cell edge length is expressed as a function of six independent variables: the effective ionic radii of the constituents (A, B and B ′ ), the electronegativities of B and B ′ , and the oxidation state of B. In this analysis, 147 perovskites of the A 2+ 2 BB ′ O 6 type, having the cubic structure and belonging to the F m3m space group, are included. They are divided in two sets; 98 compounds are used in the calibration set and 49 are used in the test set. Both models give consistent results and could be successfully used to predict the lattice cell parameter of new members of this series.