Crude distillation units are the first processing units of crude oils based on fractional distillation. The properties of the petroleum products obtained from refinery units are frequently analyzed to ensure that the off-spec product cannot be obtained and that the process is working under the desired conditions. This study aims to develop a method based on multivariate data analysis to determine physical and chemical properties of petroleum samples as an alternative to time-consuming and conventional analytical methods. Four different petroleum products obtained from CDU for years were selected and used in this study, which are heavy and light diesel, heavy and light straight run naphtha. Four different spectroscopic methods which are UV-Vis, Fluorescence, FT-NIR and FTIR-ATR spectroscopy, were performed and compared. Multivariate calibration models were developed using Partial least Squares (PLS) and Genetic Inverse Least Squares (GILS) algorithms. For heavy and light diesel, predictive performance of three different spectroscopic methods were compared and for heavy diesel UV-Vis spectroscopy, for light diesel FT-NIR spectroscopy was selected for most of the parameters. Developed models by fluorescence analysis of light diesel samples conducted with two different measurement modes and synchronized fluorescence spectral data has resulted in better models compared to total fluorescence spectra. Studies with straight run naphtha samples were obtained from three different refineries and prediction performances were compared. All obtained model results indicates that developed methodology can be used in routine operations instead of conventional analytical methods.