In this study, a hexa-β-peptide whose conformational equilibrium encompasses two different helical folds, a right-handed 2.710/12-helix and a left-handed 314-helix, is simulated using different GROMOS force-field parameter sets. When applying the recently developed GROMOS 54A7 force field, a significant destabilization effect on the 2.710/12-helix of the peptide is observed, and the agreement with the experimental NOE distance bounds is much worse compared with the ones using previous versions of the GROMOS force field. This led us to investigate the free enthalpy difference between the two helices as a function of a variation of different subsets of force-field parameters. Both long time molecular dynamics simulations and one-step perturbation predictions suggest that the disagreement with the experimental NMR data when using the 54A7 force field is caused by the use for β-peptides of the new backbone φ-/ψ-torsional-angle energy terms introduced in this force field which were based on conformational fitting of backbone φ/ψ angles for a large set of proteins. This means that these parameters of backbone φ- and ψ-torsional-angle terms should not be applied to non-α-peptides such as β-peptides. This modified assignment of torsional-angle energy terms and parameters is denoted as 54A7_β. It corrects the wrong description of the conformational ensemble of the hexa-β-peptide obtained using the previous assignment and yields as good agreement with NMR data for other β-peptides that adopt a single helical or a hairpin fold. © 2013 Wiley Periodicals, Inc.