Modern radiotherapy has increased demand for dose delivery verification. In this paper transmission portal dosimetry was considered. Portal detectors are a promising tool for 2D dosimetric verification and they are nowadays one of the most widely investigated topics. In this study an Electronic Portal Imaging Device (EPID) was positioned below the patient and the transmission images were captured during the irradiation. The principle of this verification consists of comparison of the acquired images with images predicted on the basis of the entrance fluence map and the tissue distribution in the patient. Such verification is not performed at any radiotherapy department in the Czech Republic. There is no system available for the prediction of transmission portal images. Even worldwide, there is still a lack of commercially available solutions. The aim of this paper is to present a new method of prediction of transmission portal images by means of the Monte Carlo (MC) method and the mathematical programming language MATLAB. The MC code EGSnrc (Electron Gamma Shower) was used. The validity of the presented method was verified by comparison of the predicted images with the acquired ones. The acquisition of EPID images was performed at the Hospital Na Bulovce. Three different validation tests were performed. In the first case, the EPID was irradiated by regular and irregular fields while there was nothing present in the beam path. In the second case, a water-equivalent phantom was added to the EPID and was irradiated by a number of irregular fields. In the third case, a real patient was present in the beam path and the EPID images were acquired during the patient’s treatment. The patient was irradiated by 8 treatment fields and the portal images were acquired during 5 treatment fractions. All of the acquired images were compared with the MC predicted ones by gamma analysis with gamma criteria of 3%, 3 mm. The average gamma values were 0.31−0.4, 0.34−0.4 and 0.35−0.61 in the first, second and third case, respectively. The results validate the developed method and demonstrate that MC is an effective tool for portal dose image prediction. MC may be favourably used for in vivo transmission dosimetry.