The three-dimensional distribution of the materials composing a sample is reconstructed using multiple-energy computed tomography techniques. The a priori knowledge about the specific materials that are present in the sample is used for decomposing the measured intensity projections into projected thicknesses of each component material. Those decomposed projections are then processed using iterative reconstruction algorithms to separately obtain the three-dimensional distribution of each material. A new iterative reconstruction algorithm has been implemented and its performance is compared to the standard filtered back projection and to another existing iterative algorithm based on histogram manipulation. The method is first tested with simulations on data for a realistic, three-materials" numerical sample and then applied to experimental micro-tomography data. The accuracy of the reconstruction from experimental data is tested by means of an energy dispersive x-ray spectroscopy scan of the sample.