Linear models for estimation of the first (K1), second (K2), and overall stability constant (beta2) based on the valence connectivity index of the third order ((3)chi(v)) were developed and checked on four sets of copper(II) chelates (with diamines, N-alkylated glycines, and naturally occurring amino acids, including their mixed complexes). Univariate models were valid when log K1 and log K2 values were linearly correlated, i.e., when there was no interaction between chelate rings. The univariate models proved applicable for estimation of all three stability constants of complexes with diamines and N-alkylated glycines, but for complexes with amino acids additional terms were needed (bivariate models). Models reproduced stability constants with an error usually less than 0.3 log K units.