Different simple solvent models have been implemented in an extended simulated annealing process (ESAP), developed by Higo et al. [(1992) Biopolymers, 32, 33-43] and proven to be able to predict ab initio the conformation of the antigen-combining loop H2 from FAB McPC603. The rationale used here provides a useful new method for testing solvent models in general. The different solvent models comprise a high dielectric constant, a screened coulomb potential, a dummy water model and a statistical continuum treatment of the solvent effect in which the reaction field and the solvent accessible area is accounted for. To assess the effect of the solvent, we tested the ability of simulations to retain the experimental conformation of loop H2. We compared the different structures obtained at the end of the annealing process in terms of root mean square deviation (r.m.s.d.), for both the backbone and for all atoms, root mean square (r.m.s.) fluctuation, solvent accessible surface area (ASA), hydrogen bonding network and phi-psi plot distribution. The relationship between the r.m.s.d. and the internal energy of a structure is also evaluated in terms of precision and another possible method for obtaining the best conformation is discussed. The accuracy of modelling the coarse effect of the solvent and the similarities of the resulting structures with respect to the X-ray reference structure are examined. The possible choice of one of these solvent models in the structure determination of an unknown loop structure is discussed.