To clarify the surface drift velocity in gravity waves. experimental data measured in a two-dimensional wave flume were compared with theoretical values predicted by the Stokes 2nd- and 5th- order theories as well as by the conduction solution or Longuet-Hinggins (1953). Relative water depth and wave height ranged 0.04<H/Lo<0.42 and 0.008<H/ho<0.065, respectively, in the experiment. It has been found that, as long as surface drift velocity is concerned, the Stokes 2nd-order theory shows better agreements with measured values than the Stokes 5th-order theory does, while the conduction solution shows the poorest agreement. For an open flume condition, Stokes 2nd-order theory gives higher values of surface drift velocity than the measured values for relative water depth H/Lo<0.13, while the opposite is found for H/Lo>0.13. For a closed flume condition, Stokes 2nd-order theory gives lower values than the experimental data, and the differences increase as both relative water depth and wave height increase. Based on the observed data of the surface drift velocities, a modified Parabolic model of the return current velocity Profile has been suggested, which is Proved to fit better to the existing experimental data of mass transport velocity profiles in a closed wave flume than the models of Longuet-Hinggins (1953) and Stokes wave theories do.