Photoacoustic imaging is a hybrid modality that is used to image biological tissues. Using multispectral optical excitation, a functional image is obtained due to the tissue specific optical absorption that depends on the wavelengths. To classify multispectral photoacoustic images, supervised methods are classically used. However, definition of the reference spectra is often difficult, and this choice can have a large impact on the classification results. A possible approach to build relevant reference spectra is to use spectrophotometry. This study aims at comparing absorbance measured by a spectrophotometer and multispectral photoacoustic signals of various coloured phantoms. We compare qualitatively the shape of the spectra obtained, using these two modalities for each sample. Our data suggest that spectrophotometry is a promising way to define reference spectra for classification of multispectral photoacoustic datasets.