A semi-natural grassland in Sweden was exposed to an elevated CO2 concentration during a six-year open-top chamber experiment. Vegetation composition was assessed twice a year using the point-intercept method. The field had been grazed previously, but when the experiment started this was replaced with a cutting regime with one cut (down to ground level) each year in early August. From the third to the sixth year of the study the harvested material was divided into legumes, non-leguminous forbs and grasses, dried and weighed.Elevated CO2 had an effect on species composition (as analysed by Principal Component Analysis) that increased over time. It also tended to increase diversity (Shannon index) in summer, but reduce it in spring. However, the effects of the weather and/or time on species composition and diversity were much more prominent than CO2 effects. Since the weather was largely directional over time (from dry to wet), with the exception of the fifth year, it was difficult to distinguish between weather effects and changes caused by a changed management regime.In all treatments, grasses increased over time in both mass and point-intercept measurements, whereas non-leguminous forbs decreased in mass, but not in point-intercept measurements. Legumes increased in the point-intercept measurements, but not in biomass, at elevated CO2, but not in the other treatments.Overall, we found that elevated CO2 affected species composition; however, it was only one of many factors and a rather weak one.Nomenclature: Krok & Almquist (2001).