This paper describes the development of a protocol for SPECT-based tumor dosimetry for (131)I-mIBG therapy patients with high-risk neuroblastoma. The treatment aims to deliver a whole-body dose of 4 Gy in two fractions. Whole-body retention measurements taken during the first fraction are used to guide the second therapy administration. The tumor dose from 3 patients was assessed by acquiring a minimum of three SPECT scans. Dead-time and triple-energy window scatter corrections were applied. The images were reconstructed using filtered backprojection with a Chang attenuation correction, and a phantom-based calibration factor was used to convert to activity. A monoexponential fit was made to the data, and instantaneous uptake was assumed. Tumor absorbed-dose ratios were used to analyze intrapatient variations, and absolute tumor dosimetry was used to assess interpatient variation. The whole-body dose administered ranged from (3.7 +/- 0.1) Gy to (3.9 +/- 0.3) Gy. This method is more accurate than a weight-based administration method. Despite this, a variation in absorbed tumor dose of 10-103 Gy was observed. All repeat doses were in the same order of magnitude, although 2 patients received a lower tumor dose per MBq from the second therapy owing to a shorter biological half-life. The tumor dosimetry protocol was simple to apply and reproducible, but the errors in image quantitation needed to be evaluated.