We present a comparison of trends in total column ozone from 10 two-dimensional and 4 three-dimensional models and solar backscatter ultraviolet–2 (SBUV/2) satellite observations from the period 1979–2003. Trends for the past (1979–2000), the recent 7 years (1996–2003), and the future (2000–2050) are compared. We have analyzed the data using both simple linear trends and linear trends derived with a hockey stick method including a turnaround point in 1996. If the last 7 years, 1996–2003, are analyzed in isolation, the SBUV/2 observations show no increase in ozone, and most of the models predict continued depletion, although at a lesser rate. In sharp contrast to this, the recent data show positive trends for the Northern and the Southern Hemispheres if the hockey stick method with a turnaround point in 1996 is employed for the models and observations. The analysis shows that the observed positive trends in both hemispheres in the recent 7-year period are much larger than what is predicted by the models. The trends derived with the hockey stick method are very dependent on the values just before the turnaround point. The analysis of the recent data therefore depends greatly on these years being representative of the overall trend. Most models underestimate the past trends at middle and high latitudes. This is particularly pronounced in the Northern Hemisphere. Quantitatively, there is much disagreement among the models concerning future trends. However, the models agree that future trends are expected to be positive and less than half the magnitude of the past downward trends. Examination of the model projections shows that there is virtually no correlation between the past and future trends from the individual models.