Advances in superselective microcatheterization techniques, which took place in the past decade, established superselective endovascular exploration as an integral and indispensable tool in the pretherapeutic evaluation of brain AVMs. The strict and routine application of superselective angiography furthered our knowledge on the angioarchitecture of brain AVMs, including vascular composition of the nidus, types of feeding arteries and types and patterns of venous drainage. In addition, various types of weak angioarchitectural elements, such as flow-related aneurysms, intranidal vascular cavities and varix formation proximal to high-grade stenosis of draining veins, could be identified as factors predisposing for AVM rupture. A wide spectrum of secondary angiomorphological changes induced by the arteriovenous shunt of the nidus and occurring up- and downstream of the nidus have been identified as manifestations of high-flow angiopathy. These data help to better predict the natural history, understand the widely variable clinical presentation and to define therapeutic targets of brain AVMs. Correlation of the topography of the AVM as demonstrated by MR with the angioarchitecture as demonstrated by superselective angiography provided a system for topographic-vascular classification of brain AVMs, which proved very useful for patient selection and definition of therapeutic goals. This study showed, that 40% of patients with brain AVMs can be cured by embolization alone with a severe morbidity of 1.3% and a mortality of 1.3%. Part of theses patients can, however, be cured equally effective by microsurgery or radiosurgery. Which modality will be chosen for a particular patient will mainly depend on the locally available expertise and experience, but also on the preference of the patient following its comprehensive information about the chances for cure and the risks associated with each of these therapeutic modalities. Embolization has a significant role in the multimodality treatment of brain AVMs, by either enabling or facilitating subsequent microsurgical or radiosurgical treatment. Appropriately targeted embolization in otherwise untreatable AVMs represents a reasonable form of palliative treatment of either ameliorating the clinical condition of the patient or reducing the potential risk of hemorrhage. Regarding the practical aspects of the endovascular treatment the following conclusions could be drawn from the experience obtained with this series of 387 patients with a brain AVM: (1) The goal of endovascular treatment should be defined prior to the procedure. This does not preclude a change in the goal, if additional information obtained during the procedure make this necessary. (2) The result of endovascular treatment of a brain AVM in terms of the degree of obliteration achieved and complication rate depends mainly on the endovascular strategy developed and the technique applied. These depend on the specific angioarchitecture and topography of the individual AVM, on the past history and clinical presentation of the patient and on the predefined goal of embolization. The strategy should include the definition of embolization targets, the selection of the most appropriate approach for endovascular navigation, the determination of the sequence of catheterization of individual feeding arteries, the selection of the type of catheters and microcatheters, the selection of the appropriate embolic materials as well as the site and mode of their delivery. Thereafter, every endovascular move should be, as in a chess game, the result of a logical plan. (3) Atraumatic superselective microcatheterization is a key point in the endovascular treatment of brain AVMs. It requires manual skills, knowledge of anatomy and respect for the vascular wall. (4) All locations of brain AVMs should be regarded as eloquent, and no distinction should be made between eloquent and non-eloquent areas of the brain when deciding on the execution of embolizatio