The quality of flour and its composition are essential questions for bakeries and customers. In this study, 44 different cereals were studied with colorimetry. The color indices CIE L∗, a∗ and b∗ along with the reflection spectra were measured and evaluated using multivariate statistical methods. Principal Component Analysis (PCA) and Non-Negative Matrix Factorization (NMF) were applied to characterize reflection functions and reduce the dimensionality of these data. The aim of the study was to determine whether color measurements could differentiate between whole-grain and non-whole-grain flours, and between wheat and non-wheat (triticale and rye) flours. Cluster analysis was performed on the reflection spectral data, L∗a∗b∗ coordinates, PCA coefficients, and NMF weights to identify distinct sample groups. Both dimensionality reduction methods revealed that the wheat samples studied form a four-dimensional subspace in the original multidimensional reflection spectra dataset. Furthermore, the findings confirmed that the applied methods effectively distinguish whole grain from non-whole grain flours and wheat from non-wheat flours.