Inter-ELM and ELM divertor power loads were measured on JET in dedicated deuterium, hydrogen and helium discharges. Matched triplets (D, H, He) were obtained for different values of magnetic field, B, plasma current, Ip, line average plasma density, n, and heating power, P. In this paper, the above experiments are described and the results are presented in terms of empirical scalings of inter-ELM and ELM wetted areas (power widths) versus engineering parameters. The inter-ELM wetted area on the outer target is found to scale roughly as , where A and Z and the fuel ion mass and charge numbers, and qcyl is the cylindrical safety factor, and the ELM wetted area as . The obtained inter-ELM scalings are then compared with those previously reported in the literature and with a wide range of 0D theoretical predictions. For this purpose a family of scrape-off layer power width models was constructed based on a permutation of different assumptions for parallel and perpendicular transport. It is found that a combination of parallel electron conduction and drift-ordered radial convection offers the best overall match to the empirical data, closely followed by models based on marginal stability to interchange/ballooning modes and ion convection with transport ordered radial velocity. Finally, implications for ITER are tentatively drawn, and a revised estimate for the power width in ITER is proposed. Extrapolating to ITER based on the empirical JET scaling and the optimum size scaling of R0.7±0.6, obtained based on comparison with simple models, yields a median outer target inter-ELM power width of ∼5.5 ± 2 mm (mapped to the outer mid-plane), in close agreement with previous estimates and the ITER design value. The most pessimistic forecasts (little or no size scaling and inverse linear B dependence, i.e. pure inverse current scaling) yield values as small as 2 mm. The resolution of the size scaling issue calls for additional inter-machine comparisons.