The authors' earlier theory of the quasi-biennial oscillation is reevaluated in the light of more recent observational analyses and theoretical developments. The main contention of the earlier theory, that the stratospheric oscillation is driven by the vertically propagating, short-period (5–15 days) waves excited in the upper troposphere, is now well establued. However, the assumption that the waves interact with the mean flow through critical level absorption is now seen to be largely, if not entirely, incorrect. Instead it appears that the short-period waves are strongly attenuated primarily by infrared cooling. As they are attenuated, their mean momentum flux is deposited in the mean flow. It is shown that this mechanism leads to an oscillation in the mean flow as readily as the earlier critical-level mechanism, and with fewer assumptions. Abstract The authors' earlier theory of the quasi-biennial oscillation is reevaluated in the light of more recent observational analyses and theoretical developments. The main contention of the earlier theory, that the stratospheric oscillation is driven by the vertically propagating, short-period (5–15 days) waves excited in the upper troposphere, is now well establued. However, the assumption that the waves interact with the mean flow through critical level absorption is now seen to be largely, if not entirely, incorrect. Instead it appears that the short-period waves are strongly attenuated primarily by infrared cooling. As they are attenuated, their mean momentum flux is deposited in the mean flow. It is shown that this mechanism leads to an oscillation in the mean flow as readily as the earlier critical-level mechanism, and with fewer assumptions.