The inverse folding problem of proteinlike macromolecules is studied by using a lattice Monte Carlo (MC) model in which steric specificities (nearest-neighbors constraints) are included and the hydrophobic effect is treated explicitly by considering interactions between the chain and solvent molecules. Chemical attributes and steric peculiarities of the residues are encoded in a 10-letter alphabet and a correspondent "syntax" is provided in order to write suitable sequences for the specified target structures; twenty-four target configurations, chosen in order to cover all possible values of the average contact order chi (0.2381 less than or equal to chi less than or equal to0.4947 for this system), were encoded and analyzed. The results, obtained by MC simulations, are strongly influenced by geometrical properties of the native configuration, namely chi and the relative number phi of crankshafts-type structures: For chi <0.35 the folding is deterministic, that is, the syntax is able to encode successful sequences: The system presents larger encodability, minimum sequence-target degeneracies and smaller characteristic folding time tau (f). For chi greater than or equal to0.35 the above results are not reproduced any more: The folding success is severely reduced, showing strong correlation with phi. Additionally, the existence of distinct characteristic folding times suggests that different mechanisms are acting at the same time in the folding process. The results (all obtained from the same single model, under the same "physiological conditions") resemble some general features of the folding problem, supporting the premise that the steric specificities, in association with the entropic forces (hydrophobic effect), are basic ingredients in the protein folding process. (C) 2001 American Institute of Physics.