Established methods for imaging of biological or biomimetic samples, such as fluorescence and optical microscopy, magnetic resonance imaging (MRI), X-ray tomography or positron emission tomography (PET) are currently complemented by infrared (both near-IR and mid-IR) as well as Raman spectroscopic imaging, whether it be on a microscopic or macroscopic scale. These vibrational spectroscopic techniques provide a wealth of information without a priori knowledge of either the spectral data or the composition of the sample. Infrared radiation does not harm the organism, no electric potential needs to be applied, and the measurements are not influenced by electromagnetic fields. In addition, no extrinsic labeling or staining, which may perturb the system under investigation, has to be added. The immense volume of information contained in spectroscopic images requires multivariate analysis methodologies in order to effectively mine the chemical and spatial information contained within the data as well as to analyze a time-series of images in order to reveal the origin of a chemical or biochemical process. The promise and limitations of this new analytical tool are surveyed in this review.