Three long-term experiments were carried out in meadows that were mown twice a year on three different soils: heavy clay, sand and clay-on-peat. The experiment on heavy clay studied the effect of an increase in nutrient supply on dry matter production and species diversity in a species-rich meadow. Here the increased productivity due to fertilizers reduced species diversity to only 25% of the original number of species after 30 years. In the two other experiments an attempt was made to reverse this process: to reduce annual productivity and increase species diversity by exporting nutrients through hay or sod removal. In the experimental on sandy soil, nutrient removal led to a sharp decline in production from 12 to c. 5 tonnes/ha, but nevertheless species diversity declined significantly. In the experiment on clay-on-peat the productivity declined more slowly from 12 to 7–8 tonnes/ha. The number of plant species in the plots from which the sod layer had been removed was greater than in the plots that were only mown. It is concluded that low productivity levels are essential, but cannot guarantee successful restoration of species-rich meadows.