Theoretical predictions of beauty hadron lifetimes, based on the heavy quark expansion up to and including order 1/m 2 b , do not to reproduce the experimental measurements of the lifetime ratios τ (B + )/τ (B d ) and τ (Λ b )/τ (B d ). Large corrections to these predictions come from phase-space enhanced 1/m 3 b contributions, i.e. hard spectator effects. In this paper we calculate the next-to-leading order QCD corrections to the Wilson coefficients of the local operators appearing at O(1/m 3 b ). We find that these corrections improve the agreement with the experimental data. The lifetime ratio of charged to neutral B-mesons, τ (B + )/τ (B d ), turns out to be in very good agreement with the corresponding measurement, whereas for τ (B s )/τ (B d ) and τ (Λ b )/τ (B d ) there is a residual difference at the 1σ level. We discuss, however, why the theoretical predictions are less accurate in the latter cases.