Higgs bosons with enhanced coupling to bottom quarks are copiously produced at hadron colliders via b b→ h, where the initial b quarks reside in the proton sea. We reexamine the calculation of the next-to-leading-order cross section for this process and argue that the appropriate factorization scale for the b distribution functions is approximately m h/4, rather than m h, as had been previously assumed. This greatly improves the convergence of the perturbation series, and yields a result with mild factorization-scale dependence. We also show that the leading-order calculation of g g→ b b h, integrated over the momenta of the final-state particles, is very sensitive to the factorization and renormalization scales. For scales of order m h/4 the g g→ b b h cross section is comparable to that of b b→ h, in contrast with the order-of-magnitude discrepancy between these two calculations for the scale m h. The result we obtain improves the prospects for Higgs-boson discovery at hadron colliders for large values of tan β.