We link the diagnosed model transport behavior with a model's ability to produce realistic tropical and midlatitude O3 profiles in the lower stratosphere. Models with the most realistic LS transport agree more closely with observations and each other than the models with the greatest tropical transport problems. We incorporate results from the chemistry evaluations in the SPARC CCMVal Report (2010) in order to explain the range of CCM predictions for the return to 1980 dates for global (60S-60N) and Antarctic column ozone. While Antarctic return dates are in general strongly correlated to vortex Cly levels in the LS, the correlation between vortex Cly and transport can be altered due to problems with a model's Cl chemistry or conservation. In both regions, the range of predictions made by the models with good LS transport and chemistry is smaller than the range predicted by 1) all participating CCMs , and 2) the set of CCMs that shows close agreement with observed column O3 from 1980-2006 (10 models). This study suggests that the current range of predicted return dates is unnecessarily large due to identifiable model transport deficiencies.