Photoplethysmography (PPG) is a widely used method for detecting the blood volume changes in the limbs of human body using different wavelengths of light. Many physiological parameters are estimated through the use of the PPG signals in recent studies including glycated hemoglobin (HbA1c) and blood oxygenation (SpO2). However, the processes of estimating the physiological parameters are dependent on the wavelengths of light used to acquire the PPG signal. In this study, we compare wavelengths from 300nm to 1100nm for four models to estimate HbA1c and SpO2 levels with higher accuracy. These four models were constructed from Beer-Lambert and photon diffusion theories. From the analyses of 300nm to 1100nm, it is seen that the best sets of wavelengths are found to be {426nm, 721nm, 1100nm} and {426nm, 678nm, 1100nm} for two Beer-Lambert based models, respectively. On the other hand, for both of the photon diffusion-based models, the {426nm, 384nm, 1100nm} set is found to be the best performing wavelengths. Moreover, from the analysis of the six widely used wavelengths in different PPG systems, the {465nm, 525nm, 660nm} is found to be the best performing set of wavelengths that results in higher accuracy for HbA1c and SpO2 estimation using all four models.