Traditional N fertilizer guidelines were based on relationships between N application rates and yield response. These relationships varied greatly according to residual inorganic N content in the soil. The objective of this study was to establish inorganic N threshold values for maize production on a specific soil that could also be extrapolated to similar conditions. During the 1993/94 season a site situated near Viljoenskroon (27"10'S, 26°55'E) on a sandy Clovelly soil form was selected for a NxP fertilizer trial. Five N application rates (0, 40, 80, 120 and 180 kg N ha-l ) and five P application rates (0, 10,20,40 and 60 kg P hal ) were factorially combined and replicated thrice according to a randomised block design. Treatments were repeated for five consecutive seasons on the same plots. Over seasons 100% relative yield was obtained at an inorganic N threshold value of 74 kg N ha-l , 90% at 56 kg N ha-l , 80% at 45 kg N ha-l , 70% at 35 kg N ha· l , 60% at 27 kg N hal and 50% at 20 kg N ha-l , measured approximately six weeks after planting in the 0-600 mm soil layer for the total soil volume. Values for soil volumes over rows were very similar. An average N requirement factor of 2.3 kg N applied per kg N measured was calculated for soil volumes over rows, as well as for total soil volumes, in the top 600 mm soil according to linear models. Thus by taking the inorganic N threshold values, the N requirement factor and the measured inorganic N level in the soil into account, an actual N recommendation can be calculated in order to obtain a specific relative yield between 50 and 100%. Farmers should for economic reasons be encouraged to manage inorganic N levels in the soil to obtain a certain percentage of the expected yield. Inorganic N threshold values can be extrapolated to similar soils varying in inorganic N content.