Several different frequency distributions are in common use in the frequency analysis of hydrologic events. Two two-parameter frequency distributions, the Normal and the Lognormal, are tested for best fit in the frequency analysis of available annual yield data series; and four two-parameter frequency distributions; the Normal, the Lognormal, the Gumbel, and the Log-Gumbel are tested in the frequency analysis of annual flood flow data series for 161 drainage basins in the Canadian Plains. The relative best fits are established by data transformation and subsequent linear correlation and regression analyses. The bases for comparison are the resulting correlation coefficients. It is demonstrated that of the distributions tested, the two-parameter Lognormal distribution provides the best fit for the frequency analysis of both the annual yield and annual flood flow data series for plains basins.