Monthly averaged NO2 columns obtained at twilight at 45¿S over the period December 1980 to August 1987 are presented. Improvements to the retrieval technique used previously are discussed. The slant columns presented typically range from 3 to 12¿1016 cm-2 which, because of the geometry at twilight, are approximately 20 times greater than the vertical columns. In addition to the expected diurnal and seasonal variations, the data also show variations on longer time scales. The identification of any long-term trends is masked by the effects of the El Chich¿n eruption, which was probably responsible for a decrease in the column of 1.5¿1016 cm-2 in 1983. After allowing for this, there remains a small but significant long-term trend. This trend may be due to a cooling in the stratosphere that has taken place over the observation period. Alternatively, there may be a correlation with solar activity. The observation period is too short to be able to establish a definite link, but the trend is consistent with a decrease of 0.5¿1016 cm-2 in the NO2 column from solar maximum to solar minimum. The self-consistency from year to year does not support the theory that the ''Antarctic ozone hole'' is caused by variations in solar activity which change the production of the nitrogen oxides responsible for the catalytic destruction of ozone. Both the El Chich¿n eruption and the long-term trend discussed earlier produce additive rather than multiplicative effects, with similar column reductions occurring in the summer and winter as well as in the morning and evening measurements. ¿ American Geophysical Union 1989