In this paper, the two common approaches to account for wave randomness, the spectral approach and the wave-bywave approach, are compared through numerical experiments conducted with the coupling of a surf zone hydrodynamic model and a bedload sediment transport model. Special attention is paid to the wave nonlinearity and net cross-shore bedload transport predictions. The two approaches are found to have negligible difference in their predictions of certain average hydrodynamics, such as wave heights, set-up and undertow. However, the wave-bywave approach outperforms the spectral approach in the wave nonlinearity prediction, and the two approaches differ significantly in their predictions of wave-induced net cross-shore bedload transport which strongly depends on wave nonlinearity. This suggests the necessity of using the wave-by-wave approach. The computational efficiency of the wave-by-wave approach is also discussed.