We make a brief review of the available methods to predict, in particular, secondary structures of proteins quantitatively, not only from theoretical considerations but also from various spectroscopic techniques, notably, CD, IR, Roman, and NMR. Since the quantitative data obtained on secondary structures, such as helices, sheets, turns, random coils, and unordered structures, are always compared to the available X-ray data as a validity of the method, we show from published results and from simple calculations on nearly twenty proteins that there can be substantial differences in the final results, depending on the criteria used for calculations. We point out the need for i) taking into account the solvent-induced distortion of the secondary structures, ii) carrying out complementary experiments under different physico-chemical conditions to be reasonably confident about peak assignments, and iii) calculating vibrational modes, preferably in the range of frequencies that are easily accessible to the spectroscopists. We also urge reduction of the current overemphasis of numerical analysis of spectral profiles without proper physico-chemical rationale.