Various amino acid similarity matrices have been derived using data on physicochemical properties and molecular evolution. Conformational similarity indices, CS_XX', between different residues are computed here using the distribution of the main-chain and side-chain torsion angles and the values have been used to cluster amino acids in proteins. A subset of these parameters, CS_AX' indicates the extent of similarity in the main-chain and side-chain conformations (φ,ψ and χ₁) of different residues (X) with Ala (A) and is found to have strong correlation with α-helix propensities. However, no subset of CS_XX' provides any linear relationship with β-sheet propensities, suggesting that the conformational feature favouring the location of a residue in an α-helix is different from the one favouring the β-sheet. Conformationally similar residues (close CS_AX values) have similar steric framework of the side-chain (linear/branched, aliphatic/aromatic), irrespective of the polarity or hydrophobicity. Cooperative nucleation of helix may be facile for a contiguous stretch of residues with high overall CS_AX values.