This article assesses the use of micro-Raman imaging with respect to polymer science. This relatively novel technique allows, at high spatial resolution, the acquisition of chemical and morphological information over an area of a sample. Using Raman imaging by confocal laser line scanning, a wide range of problems in polymer analysis has been studied to outline the capabilities and limitations of the technique. Three ternary polymer blends consisting of polypropene/polyethene/ethene-propene copolymer, polybutyleneterephthalate/polycarbonate/very low density polyethene, and styrene-co-acrylonitrile/styrene-co-maleicanhydrate/poly-2,6-dimethylphenylene oxide were studied with regard to compositional and morphological heterogeneities. In a binary polymer blend consisting of two different acrylate monomers, the refractive index profile established after artificially induced diffusion of the main components was determined from the concentration gradients. The distribution of unreacted free melamine in a cured melamine-formaldehyde resin was analyzed. Furthermore, the general structure of a composite sample consisting of polyethene fibers in an epoxide matrix was studied. Raman imaging proved suitable for the characterization of heterogeneities in composition and morphology on a size scale equal to or larger than 1 mu m. In this sense, the technique helps to close the gap between infrared microscopy, with its comparatively poor spatial resolution, on the one hand, and transmission electron microscopy, with its limited chemical information, on the other hand. For heterogeneities on a submicron scale, the value of the technique is limited to the determination of average information. When combined with curve fitting, Raman imaging permitted us to determine the composition of the polypropene/polyethene/ethene-propene copolymer blend with an accuracy of 5-10%. The main limitations to micro-Raman imaging of polymer systems based on the confocal laser line scanning technique have been identified as the destruction of the samples due to insufficient heat dissipation of the high-incident laser power, interferences due to fluorescence, and the stability of the instrumentation during long collection times required for good signal-to-noise ratio spectra of weak Raman scatterers.