The paper examines the dependency between total sediment transport, q, and grain size, D (i.e. q∝Dp) under dam break generated swash flows. Experiments were performed in a dam break flume over a sloping mobile sand bed with median grain sizes ranging from 0.22mm to 2.65mm. The total sediment transport was measured by truncating the flume bed and collecting the sediment transported over the edge. The experiments were designed to exclude pre-generated turbulence and pre-suspended sediment so as to focus solely on the swash flow. The magnitude and nature of the grain size dependency (i.e. p value) were inferred for different flow parameters; the initial dam depth, do, the integrated depth averaged velocity cubed, ∫u3dt, and against the predicted transport potential, qp, using the Meyer-Peter Muller (MPM) transport model and variations of that model. The data show that negative dependencies (p<0) are obtained for do and qp, whilst positive dependencies (p>0) are obtained for ∫u3dt. This indicates that a given do and qp transport less sediment as grain size increases, whereas transport increases with grain size for a given ∫u3dt. The p value is found to be narrowly ranged, 0.5 ≤ p ≤ -0.5. On average, the incorporation of a pressure gradient term via the piezometric head into the MPM formulation reduces qp by 4% (fine sand) to 18% (coarse sand). The measured total transport for fine and coarse sands is best predicted using MPM and MPM+ dp*/dx respectively. However, the inferred optimum transport coefficient in the MPM formulation is about 30, much higher than the standard coefficient in a steady flow and this is not due to the presence of the pre-suspended sediment. The optimum transport coefficient indicates some sensitivity to grain size, suggesting that some transport processes remain unaccounted for in the model.