Our approach to planning stereotactic 125I brachytherapy of brain tumors has involved least-squares optimization of individual seed positions within the target contour, followed by repeated combining of seeds from nearest-neighbor catheters in order to achieve an acceptably low number of catheters and an acceptable-separation of entry points. In one option, the catheters diverge from an extra-cranial point that can be close to the skull if all catheters are to be placed through a small craniectomy to treat a larger-diameter target. In another option, catheters converge toward a point beyond the target, to facilitate perpendicularity at the skull surface if a separate opening is to be drilled for each catheter. In either case, the fact that seed orientations are known, permits including anisotropy in dose calculations. Trial seed locations are constrained to a target region defined on a 1-mm mesh, both in the initial optimization of single-seed catheters and in subsequent combinations followed by tune-up optimizations. In the optimization process, sum-of-squares contributions are weighted more heavily when the dose rate is lower than the target dose rate; the weighting imbalance falls short of keeping all target points above the target dose rate and requires targeting on a dose rate about 25% higher than the desired minimum dose rate.