High-energy electron dose perturbations produced by variations in density and structure of the irradiated material have been demonstrated in phantoms by a number of investigators (5, 6, 10, 14–17, 19, 20), sometimes employing anatomical specimens such as bone (10, 25, 27). In one study (11), dose distortions produced by bone and air spaces in electron treatment of the nasopharynx were mapped in sectioned cadaver heads. We have investigated these dose perturbations in living tissues, using lithium fluoride thermoluminescent dosimeters. Several excellent reviews of the properties of lithium fluoride as a radiation dosimeter (7, 13) and of its use in measuring doses in irradiated patients (18, 21) and experimental animals (2) are available. In our experience, accuracies of ± 3 per cent are obtainable for lithium fluoride dosimeters in vivo, and ± 1 per cent for carefully controlled laboratory exposures. Over 500 in vivo dose determinations in 75 dogs and 25 patients comprise the data from this two-year investigation. Early results from these studies have been previously reported (3, 9, 30). Physical models of tissue heterogeneities have been employed to parallel the in vivo studies and are reported in Part II of this paper. Chest Wall A frequent clinical application of electron irradiation has been for chest wall treatment in cases of carcinoma of the breast (8, 22–24). Certain questions regarding the suitability of high-energy electrons for chest-wall irradiation required investigation: (a) To what degree is the homogeneity of irradiation of the chest wall influenced by the presence of the ribs (28)? (b) What is the effect on dose homogeneity of the variation in chest wall thickness in a given patient (28)? (c) Is there a significant loss in dose in the region of the parietal pleura due to decreased back-scattering from the underlying lung? Fifteen mongrel dogs were used for this study. At thoracotomy, lithium fluoride capsules were sutured to the inner aspect of the chest wall, alternately behind rib and in the intercostal space (Fig. 1). The row of dosimeters so placed extended craniocaudad across the middle of a 10 × 10 cm right lateral portal. The chest was then tightly closed, and intrathoracic air completely evacuated. Following closure of the chest, respiration was spontaneous. Cross-table anteroposterior and lateral localization radiographs were then taken, and the portal was irradiated with electrons of 6, 9, 12, or 15 MeV produced by a Siemens betatron. Figure 2, A and A′, demonstrate the results of such a study for 6 MeV electrons and a fairly thick chest wall. The lack of homogeneity in radiation dose at the inner aspect of the chest wall is shown graphically. The variation in relative3 depth dose is seen to be related both to chest wall thickness and to the shielding effect of the ribs.