Background: We review the accepted primary blast injury criteria. We point out a significant systematic error in the criteria due to the use of low-quality pulse duration data in the original criteria. Methods: We recalculate the injury criteria using state-of-the-art blast data. Results: We give the corrected injury criteria in the form of pressure-duration curves, pressure-impulse curves and charge weight-range curves. Conclusions. We discuss the practical implications of a naive use of the accepted criteria in the various forms. In general, the criteria based on pulse duration are seen to be over conservative, while the criteria based on impulse are seen to be too liberal.