New experimental techniques are capable of determining the relative population of conformations adopted by short alanine peptides in water. Most of the existing all-atom force fields used to model proteins fail to reproduce the relative population of the most relevant conformations of peptides. The calculated relative population of conformations varies significantly depending on the force field chosen, thus urging the need to check the validity and consistency of force fields over a range of peptide lengths. Here, we show how the applicability of a modified version of AMBER force field (A94/MOD) can extend from short to large peptides. It is also capable of reproducing the expected shift in conformational preference with increasing peptide length and temperature. Importantly, the consistency of the force field is judged by direct comparison to experiments rather than to the relative energies of conformations obtained from ab initio calculations. Importantly, this study illustrates that many aspects of protein force fields are already well refined and may only require minor refinements to accurately reproduce experimental observations over a range of systems.