In 1966, the Černíkovice experiment, Czech Republic, was started when an alluvial meadow dominated by Alopecurus pratensis was subjected to the following fertilizer treatments: non-fertilized control, PK, N50PK, N100PK, N150PK, and N200PK. The experimental plots were cut three times per year in the initial phase of the experiment and twice per year since the late 1980s. In mid May 2005, plant cover was visually estimated, biomass yield and sward height measured in order to detect changes in the grassland ecosystem caused by this long-term fertilizer treatment. After 40 years treatment was a significant predictor of sward structure, explaining 32% of the variability in plant cover data in RDA. Legumes were not detected in the N200PK treatment and achieved the highest cover in the control and PK treatment. Grasses had the lowest cover in the PK treatment and control, which significantly differed from all treatments with N. Alopecurus pratensis prevailed in all NxPK treatments. Herbs had the highest cover in the control followed by the PK treatment and both treatments significantly differed from the NxPK treatments. Achillea millefolium was recorded in all treatments, but the highest cover was recorded in the control treatment. Species richness of vascular plants ranged from 8 per m2 in the N200PK treatment to 24 in the control. A significant decrease in species richness with increase in sward height was detected. The cover of mosses ranged from 1 to 6% like sward height gradually increased with fertilizer application. Aboveground biomass yield was significantly lower in the control than all other treatments. Based on the results of the Černíkovice experiment and a comparison with other long-term fertilizer experiments it is concluded that naturally highly productive grasslands are much less threatened by the inappropriate application of fertilizers than low productive grasslands with a specific plant species composition.