OBJECTIVE Solid state diodes and/or thermoluminescent dosimeters (TLDs) are often used to measure scattered radiation doses to critical organs immediately adjacent to radiation field sites. The energy-dependent response of these commonly used in vivo dosimeters sometimes makes the interpretation of measured values uncertain. This study investigates scattered radiation arising from the collimator jaws of linear accelerators and the treatment head of a cobalt-60 teletherapy unit. METHODS A thin window Markus-type parallel-plate ionization chamber placed in a polystyrene phantom was employed to document the magnitude, energy composition, and sources of scattered radiation at surfaces near radiation fields. Measurements were taken both with and without additional phantom material covering the ionization chamber, as well as with various distances between the ionization chamber and edges of the radiation fields tested. RESULTS Data was collected, analyzed and compared for treatment units produced by different manufacturers. It was found that the magnitude of scattered radiation to surfaces immediately adjacent to radiation fields ranged from 1% to 15% of the maximum dose along the beam central axis. These values showed a strong dependence upon distance from the edge of the radiation field, beam energy, collimator setting (field size), and the presence of externally mounted accessories. Teletherapy unit differences due to manufacturing firm origins were found to only slightly affect scattered radiation magnitude, while the orientation of upper and lower collimator jaws had absolutely no effect. CONCLUSIONS Percent depth dose curves of scattered radiation were obtained and analyzed. The shapes of these depth dose curves suggest the presence of complex energy spectra from secondary electrons and scattered x-rays. Because of the presence of these complex energy spectra in areas immediately adjacent to radiation fields, caution should be observed when interpreting patient doses near radiation fields, if dose values have been measured in vivo using thermoluminescent dosimeters (TLDs) or solid state diodes. Many of these on-patient dosimetry devices are strongly energy dependent and may demonstrate large over- or under-responses in areas dominated by scattered radiation. The results of this study, thus, suggest that ionization chambers are preferred for determination of scattered radiation doses in such regions.