The quasi-biennial oscillation (QBO) in ozone in the equatorial stratosphere is obtained by analyzing the Stratospheric Aerosol and Gas Experiment (SAGE) data from 1984 to 1989. The phase of the ozone QBO in the lower stratosphere is found to precede the zonal wind QBO by several months as opposed to the theoretically expected in-phase relationship between the two. A mechanistic model is developed to explore possible reasons for this disagreement. The model is capable of simulating the actual time evolution of the ozone QBO by introducing the observed zonal wind profile as input. The modeled results confirm the conventional view that the ozone QBO is generated by the vertical ozone advection that is driven to maintain the temperature structure against radiative damping. However, a series of experiments emphasizes the importance of the feedback of the ozone QBO to the diabatic heating through the absorption of solar radiation. Due to this effect, the phase of the ozone QBO shifts up to a quarter cycle ahead and approaches that of the temperature QBO. Because of this inphase relationship, the feedback of the ozone QBO to the diabatic heating acts to compensate for the radiative damping of the temperature structure, thus reducing the magnitude of the induced diabatic circulation. Because the reduction of the magnitude of the vertical motion facilitates downward transport of easterly momentum by the mean flow, this feedback process can help to resolve the insufficiency of the easterly momentum in driving the dynamical QBO in GCMs. It should be emphasized that more sophisticated models that allow for full interaction between the chemical species and radiative and dynamical processes should be developed to improve our understanding of both dynamical and ozone QBOs.