Dosimetry of blood irradiation using x-ray beams on a medical linear accelerator has been studied to evaluate the accuracy of a diode detector and the delivery achievable. Variations in applied doses for a standard dual field 6 MV x-ray are measured with a commercial diode detector. Results show that the diode detector measured applied in vitro doses to within 5.4% (2 standard deviations (2 SD)) of those calculated with a collapsed-cone convolution treatment-planning computer for a sample of 100 blood irradiations. Experiments involving the packing procedure of the blood products in the blood box were performed. It was found that a large proportion of the variation in the predicted and measured dose was due to the compacting of the scatter material at the base of the blood box (over a 6 month period) producing a higher density below the blood than originally scanned; hence an overall reduction of delivered dose was observed. The diode measurements (which provide an immediate printout) are recommended in conjunction with a film dosimeter such as radiochromic film, which still provides a back-up dose monitor and a visual reminder that the blood has been irradiated. It is also recommended that the blood box be completely evacuated of all scatter material every month and the base be carefully repacked to provide uniform scatter material.