Development of dental caries is associated with the loss of minerals and change in the enamel structure. In this study, we have measured and compared reflectance spectra of sound and carious enamel, to investigate its utility in detection and analysis of dental caries. One hundred twenty, 3-mm diameter human enamel cores, with no sign of fluorosis, tetracycline stain, hypoplasia, fracture and restorations, were prepared. The enamel surfaces then were ground and polished. Specimens were placed on a fitted holder with either black or white color for background, with no fluorescence. The baseline spectra were measured using a spectrophotometer with enclosed diffused illumination. Spectra measured from 380 to 780 nm at 5 nm intervals. All measurements were corrected to compensate for the spectrum of illumination. The specimens were divided into two groups and exposed to a demineralizing solution, for 48 and 96 hours, respectively. Reflectance spectra of specimens were measured following lesion induction. All specimens were sectioned and analyzed by transverse microradiography (TMR), where lesion depth and mineral loss ((Delta) Z) were measured. Dimensionality of multi-spectral data was reduced through its conversion to L*a*b* color coordinates and principal component analysis (PCA). Multiple linear regression analysis showed low correlation between L*a*b* and lesion depth and mineral loss. PCA analysis showed higher correlation coefficient, compared to L*a*b*. Preliminary results of this study suggest that multi-spectral measurement and analysis of the tooth surface could be useful in predicting the depth and severity of an early carious lesion.