A novel spectral fingerprint to discriminate different dry red wines was built using data visualization method. Twelve red wines with different vintages, cultivars and ageing methods from Changli and Shacheng were sampled. Nine fractions of each wine were collected with a reversed-phase C18 column, and then they were lyophilized. The residue of each fraction was resolved with synthetic wine of the same volume with the fraction sample. The transmittance spectra of wines and their fractions were recorded from 190 to 1 100 nm. And the spectral data were visualized to show their visual differences directly. Mono-phenols in wine and fractions were analyzed by HPLC-DAD at wavelengths in the range where located the obvious differences of the spectral fingerprints. The results showed that the spectral differences of wine samples lied in the range of 190 to 600 nm. There were obvious differences in visual maps among wines with different vintages, mainly around 520 nm. The visualization differences among wines with distinct geographical origins lay in the F8 maps, and the differences from the aging methods almost cover the whole wavelength range visualized. However, wines from different grape cultivars had the similar visual characteristics. HPLC-DAD identified the possible mono-phenol groups for the spectral differences at 280, 313, 365 and 520 nm. It was concluded that the visualization of spectral data from 190 to 600 nm could be used to build red wine spectral fingerprint to distinguish dry red wines with different vintages, origins, and ageing methods.