SUMMARYNineteen NP experiments were carried out with wheat and barley mostly in the Mesaoria plain of Cyprus over the period 1968–78. The fields were fallowed over the season preceding the experiments. Rainfall ranged from 68 to 405 mm. Kyperounda, a tall durum wheat, Pitic 62, a semi-dwarf aestivum wheat, and Athenais barley were the test varieties. No yield was obtained when rainfall was below 100 mm. Athenais barley consistently outyielded the wheat varieties, particularly Kyperounda, by up to 200% presumably because of earlier heading, which helpedthe crop to escape drought to a certain extent.Kyperounda wheat and Athenais barley mostly responded to 35 kg N/ha but Pitic 62 benefited from up to 70 kg N/ha. Fertilizer N not taken up in a year of extremely low rainfall remained in the soil and was available to the following crop.Response to P was more marked when rainfall was below 250 mm being linear over the rates tested (highest rate 26 kg P/ha). Athenais barley responded to P more than the other varieties. Bicarbonate-soluble soil P was between 1 and 10 mg/kg.Since rainfall is unpredictable it is recommended that 20–40 kg N/ha be applied atseeding together with 13–26 kg P/ha. A similar amount of N should be top dressed in late January. The higher amount refers to the semi-dwarf Pitic 62 wheat and could be increased further if the December-January rainfall, which normally constitutes 40% of the total, is high. These rates refer to crops grown after fallow. For continuous growing the rates would probably have to be increased.