Raman optical activity (ROA) spectroscopy combined with quantum-chemical simulations is a sensitive method to determine the absolute configuration and conformation of chiral molecules in solutions. However, the precision of this approach varies for different systems. In the present study, the reliability and numerical stability of decomposing experimental spectra into calculated subspectra is tested on the Ala-Ala dipeptide. Molecular dynamics (MD) snapshots of Ala-Ala/water clusters are averaged to account for solvent effects and molecular flexibility. Multiple experiments with protonated, zwitterionic, and deprotonated dipeptide forms and natural and d(2)- and d(8)-isotopically labeled dipeptides are used to verify the results and estimate the overall accuracy. Although the precision is still limited by experimental noise and computational error, a very close match between the observed and theoretical spectral shapes has been achieved. This enabled quantitative determination of conformer populations with a typical dispersion of 10%. The spectroscopy also demonstrated how the conformation depends on pH. The ROA results were more consistent than the Raman ones. Typically, the ROA analysis was more resistant to artifacts in the experiment, such as incomplete baseline subtraction. Conformer ratios predicted by MD agree fairly but not fully with the experimental ones. This indicates minor deficiencies in the Amber force field, particularly for the protonated dipeptide. Overall, the combination of ROA experiment and computational chemistry appears to be a robust tool providing deep insight into molecular structure.