In this article, the submersion dose due to a radioactive cloud of pollutants was evaluated at short downwind distances from an emission stack. The atmospheric transport of contaminants was modelled using the Gaussian plume model (GPM). The algorithm for dose computation and its hypotheses were analysed. Two relevant issues were discussed: the semi-infinite cloud approximation used for pre-calculated dose conversion factors and the lack of a radiation transport model for dose computation outside the radioactive cloud. The GPM-based software HotSpot and GENII V2.10 and a FLUKA Monte Carlo GPM implementation were compared in a scenario characterized by a low release height and two different simplified atmospheric conditions. Compared to FLUKA, HotSpot and GENII V2.10 results showed a significant dose overestimation inside the plume. Moreover, in extremely stable meteorological conditions, only the Monte Carlo code could detect the ground-level dose contribution from an overhead plume.