The radial electric field in known to be one of the drivers of parallel ion flow in the SOL. It contributes to the ion Pfirsch-Schluter flow and also determines the 'return parallel flow' that can arise to compensate poloidal E x B drift. It was established recently that 2D fluid codes EDGE2D and SOLPS underestimate the predicted E, in the SOL compared to experimentally measured values. The underestimate is likely to be related with kinetic effects of parallel transport by heat-carrying electrons. The present work demonstrates that this underestimate can be responsible for the large discrepancy between measured and simulated parallel ion flows in the SOL observed in a number of experiments. Were the radial electric field modelled correctly by the codes, a significant increase in the predicted Mach number of the parallel ion flow could be expected. For the part of the ion flow that depends on the toroidal field direction, this would greatly reduce, or even possibly eliminate, the difference between the experiment and model, as concluded from the comparison between measured and simulated radial electric fields and Mach numbers of the parallel ion flow in JET and ASDEX Upgrade.