Examination of radial electric field (E,.) profiles in the scrape-off layer (SOL) of ASDEX Upgrade (AUG) and JET revealed large discrepancies between 2D fluid edge modelling and experiment. Experimental profiles of plasma potential (V-p) in the outer (low field) side of the plasma, obtained with reciprocating Langmuir probes, decay radially with electron temperature, T-e, with the -eE(r)/del T-e ratio being > 1.5. In contrast, code simulated E-r are fairly low in most of the SOL (compared with -del T-e/e). Modelling with kinetic treatment of neutrals and drifts was performed using the SOLPS code for AUG cases and EDGE2D-Nimbus for JET cases. Mismatches between modelled and experimental E-r may be caused by the recently established tendency for the SOLPS code to underestimate T-e in the divertor of AUG. It was attributed to non-locality of parallel transport of supra-thermal, heat-carrying electrons originating upstream of the divertor, which are usually only weakly collisional and can penetrate, with few collisions, to the target. Ratios -eE(r)/del T-e obtained from the probe measurements in JET are of order 1.6, while in AUG, JT-60U and TCV they are of order 3. Such high values point to the possibility of fast electrons contributing, apart from target heat fluxes, also to the formation of the Debye sheath. The problem of the underestimation of E-r. in the codes must be closely related with the well-known problem of the underestimation of those parts of parallel ion flows in the SOL that are influenced by the toroidal field direction. It was demonstrated earlier that parallel ion flow at the outer midplane is dominated by the ion Pfirsch-Schluter flow, which in turn is partly driven by the radial electric field. The T-e and E-r discrepancies, as well as discrepancies between simulated and experimental parallel ion flows, raise a question of the validity of fluid codes for the plasma edge modelling and prompt the inclusion of kinetic effects into present-day 2D fluid codes which assume strong collisionality.