The emergency of interventional revascularization techniques in the treatment of atheromateous vascular diseases has resulted in the need for additional valuable diagnostic information about the 3D morphology and nature of the lesion. To overcome the limitations inherent to common vascular imaging techniques, such as Digital Angiography (DA) which only gives partial information on lumen narrowing or Intravascular Ultrasound (IVUS) which provides randomly oriented transversal images, a 3D reconstruction of the vessel by fusion of X-ray and IVUS images has been developed. For that purpose, X-ray and IVUS images are acquired according to a well-defined protocol and useful information to be fused is extracted. A geometric model then leads to the determination of the unknown parameters which allow the alignment of all data in a common reference frame. The registered data are then directly introduced into a probabilistic reconstruction process using a Markovian modeling associated with a simulated annealing-based optimization algorithm. Taking into account all the information available about the vessel, the method avoids the uncertainties and ambiguities of a reconstruction based only on one modality, and the probabilistic fusion solves the possible contradictions between both acquisitions. Results of vascular lumen 3D reconstruction are shown with data acquired on an excised dog aorta. The accuracy of reconstruction of the lumen by data fusion is significantly improved compared to results obtained with separate reconstruction from angiographic or ultrasonic data. Further work will include introduction of vessel wall texture elements into the probabilistic fusion process to increase the amount of information gained by intravascular ultrasonic tissue characterization.