A three-dimensional offline transport model is used to study the evolution of the Antarctic ozone hole in spring 1994. The modelled ozone profiles generated using winds from the UK Meteorological Office (UKMO), including the whole stratosphere, are closer to observations than simulations driven by operational weather analyses from the Global Assimilation and Prediction system (GASP), with highest level at 10 hPa. Model simulations are also fairly insensitive to horizontal or temporal resolution of the input driving fields, particularly at middle and high latitudes. While the large-scale synoptic variations in simulated total ozone are in good agreement with satellite observations, the model underpredicts total ozone at high latitudes during the ozone hole recovery phase. Further analysis suggests that the polar stratospheric clouds (PSC) chemical parametrisation is not appropriate for the specific spring of 1994. In addition, our results also suggest that dynamics are partially responsible for the slower than observed recovery of column ozone.