Reconstruction of vascular trees from digital diagnostic images is a challenging task in the development of tools for simulation and procedural planning for clinical use. Improvements in quality and resolution of acquisition modalities are constantly increasing the fields of application of computer assisted techniques for vascular modeling and a lot of Computer Vision and Computer Graphics research groups are currently active in the field, developing methodologies, algorithms and software prototypes able to recover models of branches of human vascular system from different kinds of input images. Reconstruction methods can be extremely different according to image type, accuracy requirements and level of automation. Some technologies have been validated and are available on medical workstation, others have still to be validated in clinical environments. It is difficult, therefore, to give a complete overview of the different approach used and results obtained, this paper just presents a short review including some examples of the principal reconstruction approaches proposed for vascular reconstruction, showing also the contribution given to the field by the Medical Application Area of CRS4, where methods to recover vascular models have been implemented and used for blood flow analysis, quantitative diagnosis and surgical planning tools based on Virtual Reality.