PURPOSE: To assess the accuracy of transmission dose rate measurements for various phantom-detector geometries, performed with an electronic portal imaging device (EPID) and to compare these transmission dose rate values with exit dose rate data. METHODS AND MATERIALS: Transmission dose rate values on the central beam axis and beam profiles were measured with an EPID consisting of a matrix of liquid-filled ionization chambers. These data were compared with transmission and exit dose rate values, obtained using air-filled ionization chambers for a number of field sizes, phantom thickness, and phantom-detector distances. Various homogeneous and inhomogeneous phantoms were applied. RESULTS: The increase in dose rate with field size is larger for the EPID than in air, due to the larger amount of side scatter in the EPID. The difference has been taken into account by a deconvolution of the EPID images. An additional build-up layer on top of the commercial device is needed to reach dose maximum at the liquid ionization chambers for photon beam energies higher than about 4 MV. The transmission off-axis ratios (OAR) determined with the EPID and in air agreed within 2% for all tested cases, after deconvolution of the EPID signal. The agreement between the EPID-and exit-OAR decreased with increasing phantom-detector distance and the presence of inhomogeneities. For a phantom-detector distance of about 10 cm, the EPID- and exit-OARs agree within 2.5%. The difference could be up to 8% for an air inhomogeneity and a phantom-detector distance of 30 cm. CONCLUSIONS: The difference between EPID measurements and measurements in air can be explained by side scatter effects in the EPID and lack of adequate buildup, and can easily be taken into account. The loss of scatter compared with the situation at the exit side of the phantom explains the difference between transmission and exit dose values. At short phantom-detector distances, good agreement exists between transmission and exit dose rate. This implies that at this distance, the EPID can be used for simple comparison with exit dose calculations during patient treatments. At larger distances, more sophisticated conversion methods are required.