[en] The constraints of fixed-target DIS data in fits of parton distributions including QCD corrections to next-to-next-to leading order are studied. We point out a potential problem in the analysis of the NMC data which can lead to inconsistencies in the extracted value for α_s(M_Z) and the gluon distribution at higher orders in QCD. The implications for predictions of rates for Standard Model Higgs boson production at hadron colliders are investigated. We conclude that the current range of excluded Higgs boson masses at the Tevatron appears to be much too large. (orig.)