We calculate the branching ratio for B --> J/psi+anything, within the
color-singlet approximation for J/psi production, but including pertur
bative QCD corrections beyond the leading logarithm approximation. Suc
h higher order corrections are necessary, in order to obtain a result
that is not strongly dependent on the renormalization scale. As in the
earlier work of Bergstrom and Ernstrom, we use a double expansion in
alpha(s) and in the small ratio of Wilson coefficients L-0/L-2, to ide
ntify the dominant terms in the decay amplitude. We complete their wor
k by calculating all the leading order terms in this double expansion.
The predicted branching ratio is then B(B-->J/psi+anything) = 0.9(-0.
3)(+1.1) x 10(-3), which is well below the experimental value B-expt =
(0.80 +/- 0.08)%. This confirms the suspicion that nonpertubative cor
rections to the color-singlet approximation for J/psi production in B
decays are important.