The relationship between the grain size distribution of the sediment on the bed and that found in suspension due to wave action above ripples is assessed here using detailed, pumped sample, measurements obtained at full-scale and also at laboratory scale. The waves were regular and weakly asymmetrical in most tests, and irregular in a minority of tests. The beds comprised fine and medium sand and were rippled in all tests. The cycle-mean sediment concentrations (C) from the pumped samples were split into multiple grain size fractions and then represented by exponential Cprofile shapes. The analysis of these profiles was carried out in two stages to determine: i) the relationship between the size distribution of the sediment on the bed and that found in the reference concentration, and ii) the behaviour of the exponential decay scale of the C-profiles. From this analysis inferences are made about the relative roles of diffusion and convection in the upward sediment flux linked to the process of vortex shedding from the ripple crests. The Transfer Function (Tr) defined to relate the bed sediment size distribution to that of the reference concentration indicates that, while finer fractions are relatively easily entrained, the suspension of some coarser fractions is caused by an additional convective effect that supplements diffusion. The evidence for this becomes pronounced above steep ripples, and the Transfer function suggests further that irregular waves increase the occurrence of coarser fractions in suspension. A functional form for Tr is suggested incorporating these principles. The exponential decay scale LS arising from the fractional C-profiles is also examined to assess the mechanisms responsible for the upward transfer of grains and a parameterisation of LS related to ripple size is suggested. The separate findings for Tr and LS present supporting evidence of diffusion affecting the finer fractions in suspension and combined diffusion + convection affecting the coarser fractions. The methodology developed allows the vertical profile of suspended median grain size to be predicted given knowledge of both the bed grain size distribution and also the flow conditions.