Stereotactic arc treatment of small intracranial tumors is usually performed with arcs collimated by circular cones, resulting in treatment volumes which are basically spherical. For nonspherical lesions this results in a suboptimal dose distribution. Multiple isocenters may improve the dose conformity for these lesions, at the cost of large overdosages in the target volume. To achieve improved dose conformity as well as dose homogeneity, the linac jaws (with a minimum distance of 1.0 cm to the central beam axis) can routinely be used to block part of the circular beams. The purpose of this study was to investigate the feasibility of blocking cones with diameters as small as 1.0 cm and a minimum distance between the jaw and the central beam axis of 0.3 cm. First, the reproducibility in jaw positioning and resulting dose delivery on the treatment unit were assessed. Second, the accuracy of the TPS dose calculation for these small fields was established. Finally, clinically applied treatment plans using nonblocked cones were compared with plans using the partially blocked cones for several treatment sites. The reproducibility in dose delivery on our Varian Clinac 2300 C/D machines on the central beam axis is 0.8% (1 SD). The accuracy of the treatment planning system dose calculation algorithm is critically dependent on the used fits for the penumbra and the phantom scatter. The average deviation of calculated from measured dose on the central beam axis is -1.0%+/-1.4% (1 SD), which is clinically acceptable. Partial cone blocking results in improved dose distributions for elongated tumors, such as vestibular schwannoma and uveal melanoma. Multiple isocenters may be avoided. The technique is easy to implement and requires no additional workload.