The recombination of methyl radicals is the major loss process for methyl in the atmospheres of Saturn and Neptune. The serious disagreement between observed and calculated levels of CH3 has led to suggestions that the atmospheric models greatly underestimated the loss of CH3 due to poor knowledge of the rate of the reaction CH3 + CH3 + M - C2H6 + M at the low temperatures and pressures of these atmospheric systems. In an attempt to resolve this problem, the absolute rate constant for the self-reaction of CH3 has been measured using the discharge-flow kinetic technique coupled to mass spectrometric detection at T = 202 and 298 K and P = 0.6-2.0 Torr nominal pressure (He). CH3 was produced by the reaction of F with CH4, with [CH4] in large excess over [F], and detected by low energy (11 eV) electron impact ionization at m/Z = 15. The results were obtained by graphical analysis of plots of the reciprocal of the CH3 signal vs reaction time. At T = 298 K, k(1)(0.6 Torr) = (2.15 +/- 0.42) x 10(-11) cm(3) molecule(-1) s(-1) and k(1)(I Tort) = (2.44 +/- 0.52) x 10(-11) cm(3) molecule(-1) s(-1). At T = 202 K, the rate constant increased from k(1)(0.6 Torr) = (5.04 +/- 1.15) x 10(-11) cm(3) molecule(-1) s(-1) to k(1)(1.0 Torr) = (5.25 +/- 1.43) x 10(-11) cm(3) molecule(-1) s(-1) to k(1)(2.0 Torr) = (6.52 +/- 1.54) x 10(-11) cm(3) molecule(-1) s(-1), indicating that the reaction is in the falloff region. Klippenstein and Harding had previously calculated rate constant falloff curves for this self-reaction in At buffer gas. Transforming these results for a He buffer gas suggest little change in the energy removal per collision, -(d), With decreasing temperature and also indicate that - (d) for He buffer gas is approximately half of that for Argon. Since the experimental results seem to at least partially affirm the validity of the Klippenstein and Harding calculations, we suggest that, in atmospheric models of the outer planets, use of the theoretical results for k(1) is preferable to extrapolation of laboratory data to pressures and temperatures well beyond the range of the experiments.