This experiment was carried out to investigate the effect of pure mineral nitrogen fertilization on dry matter yield and to determine the amounts of advisable mineral nitrogen according to difference of annual precipitations in permanent grassland. The results obtained were as follows: 1. Only PK-fertilizing without additional mineral nitrogen application produced 6. 9 tonlha of annual dry matter yields on average in relatively wet years that was higher than in rel. dry and normal years under 3-cut system by 0. 7 and 0.6 ton DMIha, respectively. 2. Due to the lesser rainfall the nitrogen-efficiency was substantially higher with heavy dressing of nitrogen fertilizer. The absolute differences in yield between rel. dry and wet years were 0.4 and 0.7 ton DMIhalyear respectively when dressed with 90 and 120 kgN/ha/cut at 4-, 5- and 6-cut systems, whereas at 3-cut system differences around 1.3 and 1.1 ton Dhllhalyear respectively were recorded. 3. In rel. dry years the most efficient N-dressing rates per ha and year tended to be slightly higher than in rel. wet and normal years. Particularly at 5-cut system 4-7 kg/ha/cut of more nitrogen fertilization was required in order to obtain the highest overyields. 4. The N-dressing rates needed to maintain a nitrogen-efficiency of 8 to 16 kg DM/kg N tended to be stronger particularly at high cut system, and also in rel. dry years higher dressing rates were required that in rel. wet and normal years.