Variations in the 18-year (1971--1988) Hohenpeissenberg (Federal Republic of Germany) and the 33-year (1956--1988) Arkona (German Democratic Republic) surface ozone data sets are analyzed. Over the full period of the records (as made available to us), both stations, which are about 800 km apart, show a positive long-term trend of about 1.0% per year (strongest in winter at Hohenpeissenberg and in spring at Arkona). There are, however, marked differences in the fluctuations over various subperiods at the two locations. For example, the Hohenpeissenberg data show a larger increase in mean surface ozone concentration in the 1970s (about 2.1% per year, strongest in winter) compared with that in the 1980s (about 0.5% per year, strongest in summer). The Arkona data show a significant linear increase from 1956 to 1979 (about 2.4% per year, strongest in winter in the 1960s and in summer in the 1970s) but a linear decrease in the 1980s (about -2.4% per year, strongest in winter) and over the period 1971--1988 (about -1.2% per year, strongest in autumn and weakest in spring). The decrease is caused by the considerably lower concentrations in the 1980--1985 period when concentrations declined at about -10% per year. Over the same period 1971--1988, the seasonal cycle at Hohenpeissenberg exhibits a summer maximum in July (with a broad peak), while that at Arkona displays a spring maximum in May (with a sharp peak). The causes of these differences are likely to be complex: a combination of photochemistry (which depends on the distribution of precursors, particularly NOx), differences in surface deposition, local departures in the atmospheric circulation and, possibly, data quality. The pre-1976 Hohenpeissenberg data and the pre-1972 Arkona data (before filters were used to remove SO2) and the post-1982 Arkona data (after a new measuring instrument was installed) may need to be further scrutinized to ensure consistency in data quality. ¿ American Geophysical Union 1990