The temperature- and pressure-dependent behavior of the cross section for optical absorption by the methyl radical is carefully considered, so we may define a criterion for selecting and correcting measurements of the rate coefficient for the recombination of methyl radicals, CH3 + CH3 --> C2H6. The low-temperature data of Slagle et al., Hippler et al., and Waiter et al. and the high-temperature data of Glanzer et al., Hwang et al., and our latest results (previous paper in this issue) are used to define a data set which contains 217 points. Subsets of isothermal data show that the temperature dependence of the high-pressure rate coefficient may be described by the simple exponential function A(infinity) exp{-T(K)/T-infinity}. Four different formulations for the pressure dependent behavior in the falloff region are used for the global fits : (1) the asymmetric Lorentzian broadening function of Gilbert et al.; (2) the Gaussian broadening function of Wang and Frenklach; (3) the empirical "a equation" introduced by Gardiner; (4) the extension of Lindemann’s expression suggested by Oref. All formulations reproduce the data, but Oref’s "J equation" produces the least correlation between the best-fit parameters, the least uncertainty in these parameters, and the smallest uncertainty in the predictions. These results are k(infinity)(T) = 8.78 x 10(-11) exp{-T(K)/723} cm(3) s(-1), k(0)(T) = 9.04 x 10(-27) cm(6) s(-1), and J(exp)(T){exp[T(K)/268]-1}(2).