In the present study, we examine the spatial and seasonal variability in predicted sFe deposition from 6 CMIP5 models, all including dust sources with a fixed Fe solubility, compared with chemical transport model simulations. Three CMIP5 models (CESM1, HadGEM2 and GFDL-ESM2M) predict the largest Fe deposition in the subtropical NE Atlantic and the Arabian Sea, while more Fe is predicted by the other three models (MPI-ESM, CNRM-CM5 and IPSL-CSM5A) to deposit to the mid latitude North Pacific and the Mediterranean Sea. Sensitivity studies with a chemical transport model suggest that inclusion of dust Fe dissolution schemes and combustion sFe sources modifies the predicted sFe deposition in both magnitude and spatial distribution, and the uncertainty is comparable to the inter-model variability due to the dust sources alone. Furthermore, we show that the model differences in sFe deposition are not reflected consistently on the predicted sea-surface dissolved Fe (dFe). In addition to better agreement in the global mean dFe suggested in the previous studies, the spatial and seasonal variability in the dFe distributions across the models is shown to be decreased. Locations with high concentrations of dFe are generally not co-located with large deposition of sFe. But the extent of the dFe distribution deviated from the sFe deposition vary significantly from one model to another. This implies that response of sea-surface dFe distributions to anthropogenic influences (through dust Fe dissolution or combustion sources) depend on the model representation of Fe cycle in the ocean.