Introduction The aim of our study was to investigate the dosimetric accuracy of Analytical Anisotropic Algorithm (AAA) and Acuros XB (AXB) with the two absorbed dose reporting options dose-to-medium ( D m ) and dose-to-water ( D w ) . The calculations were carried out for small multi-leaf collimator (MLC) fields in heterogeneous media against Monte Carlo (MC) simulations and EBT3 radiochromic films measurements. The well-known approach of Siebers et al.1 used for converting D m to D w in MC codes was also investigated. Methods Two multilayered slab phantoms were used for this study. These phantoms consisted of soft tissue equivalent material with bone (1.82 g/cm3) or Lung (0.29 g/cm3) slabs. MC calculations were performed using GATE platform and were considered as benchmark for D m calculations. The AXB D m calculations were compared to those of GATE D m with the AXB library elemental compositions on one hand and with the experimental phantoms elemental compositions on the other hand. The results of the AAA, AXB D w and GATE D w . were compared against film measurements considered as D w reference. All calculations and measurements were performed for 6 MV photon beams delivered from a Truebeam linear accelerator with a 3 × 3 cm2 nominal MLC field. Results The AXB D m calculations showed excellent agreement with GATE D m when the calculations were performed with the same elemental compositions. However, an average difference of 2% and 1.6% was found between AXB D m and GATE D m with the experimental phantom compositions in Lung and bone, respectively. The GATE D w and AXB D w calculations were in agreement with films measurement with an average difference less than 1.5% for the Lung configuration. Conversely, an average differences of 3.1% was found with the AAA results after the lung region. The results of AXB and GATE calculations were significantly different compared to film measurements in the bone region. Differences were up to 10.4% and 7% for AXB D w and GATE D w , respectively. These results show that the conversion method used by AXB and GATE should be reviewed. Concerning the GATE conversion method, a fluence correction factor should be applied as mentioned by Andreo [1] . This factor was calculated for the studied media and good agreements were found with measurements after applying it. Conclusions The AXB was found to be more accurate than the AAA in the low density media. The current methods employed for converting D m to D w in AXB and GATE are invalid and should be corrected. Further studies on the calculation of electron spectra and fluence correction factors in different media and for various field sizes are ongoing.