Background: To quantify the influence of photon dose-calculation algorithm selection on the cervical esophagus (CE) dose indices and the derived equivalent uniform dose (EUD) and normal-tissue complication probability (NTCP) for acute esophagitis in patients with head-and-neck cancer (HNC). Materials and Methods: The Fast Photon Effective Path (FPEP) and Collapsed-Cone Convolution Superposition (CCCS) algorithms on the Prowess Panther treatment planning system were compared for 30 patients (six tumor sites). The Lyman-Kutcher-Burmann (LKB) model was used to calculate the EUDs and NTCPs. Results: On average, the more simplistic FPEP algorithm overestimated the mean dose to CE planning organ-at-risk volumes (PRVs) by 2.0% (p = 0.003). The average absolute difference in mean dose was 2.7% and the maximum difference was 9.3%. The V 5Gy , V 10Gy , V 15Gy , V 20Gy , V 25Gy and V 30Gy values were significantly higher with FPEP, while the point-dose and D 2cc hot spots were similar. In turn, the dose differences led to an underestimation of the LKB-model prediction of the EUD by 1.4% (p = 0.297). The mean absolute difference in EUD was 4.5% and the maximum difference was 15.3%. In the 14-50 Gy mean dose range, the resulting NTCPs with FPEP were lower on average by 2.6% than CCCS (p = 0.041). Conclusions: In the group of HNC patients considered in this study, the EUD and NTCP for acute esophagitis showed to be moderately sensitive to the choice of dose-calculation algorithm. Despite an overestimated mean dose by the simpler algorithm, the NTCP underestimation, which can be large in some patients, is of clinical concern.