We consider the recent results of Kinoshita in which he improves the accuracy of the theoretical value for the anomalous magnetic moment of the muon. This is needed now that a new, more accurate experiment has been approved at Brookhaven National Laboratory. Kinoshita's results are completely numerical. Here we perform an independent check of his results in fourth and sixth order by analytical means, using expansions in the small mass ratios which occur in the computation. Our result for the fourthorder contribution is a: ' -ai4'=5 904475.1(3)X 10-12. This is to be compared with Kinoshita's result a:4'-a64'=5 904485X 1 0 12. For the sixth-order vacuum-polarization contribution we obtain (a:' -aj6')(vacuum polarization) =24 064.8 ( 6) X lo-''. Kinoshita's result is (a:' -aj6')(vacuum polari~a t i o n ) = 2 4 0 6 9 ( 6 ) X l O -'~. Our result for the total QED contribution is a$ED = 1 165 846 943(28)(27) X lo-''. This agrees with Kinoshita's result u,U"~(K) = 1 165 846 961(44)(28) X lo-''. Our final result for the muon anomaly is ~h~' " '~ =116591901(77)X10-11. This should be compared with Kinoshita's result ahheorY = 116 591 919( 176) X lo-'' and the experimental value ayPt = 1 165 923(8.5) XOur value makes use of a recent computation of the hadronic contribution, in which the error may be overly optimistic.