Dry hill country provides a diverse environment in which variations in aspect and slope affect pasture growth and soil processes in a number of different ways. The AgResearch fertiliser and pasture growth trial at Waipawa in the east coast of the North Island of New Zealand, provided an ideal location, with contrasting topography and fertiliser treatments at which to measure pasture growth constraints, with particular reference to nitrogen (N), in dry east coast hill country over the period of a year. Initial soil testing at the trial site revealed large differences in soil total N and phosphorus (P) concentrations between aspects, slopes and P fertility regimes in the top 75 mm of the Waipawa silt loam soil. The measured differences in soil N concentrations (ranging from 2.60-5.80 mg N/g) are a result of contrasting energy and moisture regimes between the aspects and slopes, affecting soil forming processes and subsequent soil total N concentrations. The differences in soil total P concentrations (ranging from 0.42-0.88 mg P/g) between aspects and slopes can be largely attributed to stock transfer, while the variation between P fertility regimes can he attributed to a large application of P fertiliser to the high P treatment before commencement of the trial. The existing trial provided three main contrasts over which the major restrictions on pasture growth could be measured. These were phosphate status (high phosphate (HP) and low phosphate (LP)), aspect (North and South), and slope (steep (25° +), easy (15-20°) and flat (0-12°)). This gave a total of ten combination of P status and topography Each of these sites was replicated three times to give a total of 30 locations. At each location two 0.5 m2 cages were used to measure pasture growth. One of the cages was used to measure pasture growth under the current soil fertility regime, while the other cage had 120 kg N/ha as urea applied to it, to allow growth under non-limiting N conditions. The cage locations were pretrimmed before cage placement and pasture was cut monthly (or when growth conditions permitted) and analysed. Soil samples were taken from adjacent to the cage under which pasture growth was measured in the absence of added N fertiliser. Each sample comprised six cores that were taken to a depth of 75 mm at the start of each growth period and analysed for a range of soil measurements, which were related to pasture growth. There was a wide range of total pasture yield and pasture growth rates between sites and seasons. Measured total annual pasture production ranged from 2394 - 9067 kg DM/ha/yr in the absence of added N on HP northerly steep (HPNS) and HP northerly easy (HPNE) sites respectively. Total annual pasture production in the presence of non-limiting N (potential yield) ranged from 8001 - 18532 kg DM/ha/yr on LP northerly steep (LPNS) and HPNE sites respectively. Daily pasture growth rates in the absence of added N ranged from 1.13 kg DM/ha/day on the LPNS sites in mid spring to 34.51 kg DM/ha/day on HP southerly easy (HPSE) sites in the same period. Daily pasture growth rates in the presence of non-limiting N ranged from 3.78 kg DM/ha/day on LPNS sites in mid spring to 77.95 kg DM/ha/day on the HPNE sites in early spring. The major constraints on pasture growth within each site fluctuated throughout the year, between climatic and nutrient inputs. However, the most consistent and major constraint on all sites was N deficiency. Total annual pasture yield depression from potential due to N deficiency ranged from 5757 - 9465 kg DM/ha/yr on LPNS and HPNE sites. Daily pasture yield depression from potential growth rate ranged from 2.65 kg DM/ha/day on LPNS sites in mid spring to 45.36 kg DM/ha/day on HPNE sites in early spring. Increased P levels provided increased total annual yields, with HP sites yielding on average 1038 and 2209 kg DM/ha/yr more pasture in the absence and presence of non-limiting N respectively than LP sites. NE and flat sites produced the largest and most consistent P responses throughout the year, whilst the P response was generally largest in dry periods when soil moisture was limiting. This was suggested to result from an increased rooting depth on HP sites allowing them to source more water. Sunlight and temperature were major constraints on pasture growth on southerly and flat sites throughout late autumn and winter, whilst soil moisture was the major constraint on northerly sites from mid spring through to the end of summer as well as on southerly and flat sites from mid to late summer. Clover yields were small and scattered throughout the trial in all seasons. The HPNE and HPSE sites generally yielded the most clover throughout the year, with the highest yield of 1497 kg DM/ha/yr measured on HPNE sites in the presence of non-limiting N. The Mineralisable N soil test showed some potential for use as an indicator of potential pasture yield, however the varying temperature regimes between aspects and slopes limited its applicability to all sites due to the heavy dependence of the mineralisation rate on temperature. The Olsen P and ammonium-N and nitrate-N soil tests provided no significant relationship with pasture yield. A pasture growth model relating pasture growth (G) to evapotranspiration (Et) according to the equation G = kEt (where the proportionality constant (k), with units of kg D.M/ha/mm, is a site-specific factor which could be an index of soil fertility status), showed potential to be used in hill country. The model was able to account for much of the greater than two fold difference in total annual pasture yields between the lowest producing and highest producing sites in the presence of non-limiting N (R2 = 0.66), using a single growth constant (k) for LP and HP sites. When separate k values were used for LP and HP sites the relationship between measured and predicted total annual yield improved (R2 = 0.89 & 0.73 for LP and HP sites respectively). When the seasonal accuracy of the model was tested, prediction on some sites in some periods was found to be inaccurate. Some simple adjustments such as altering the depth of available water and winter growth rates on southerly steep (SS) sites was able to increase the accuracy of the model on some of the sites. Major factors decreasing the accuracy of the model appear to be depth of available water, soil hydrophobicity in summer, recharging of soil water through capillary action in spring and warm convection currents increasing pasture growth on SS sites in winter to above predicted levels. Further development of the model will require the effect of these factors to be quantified and accounted for. The model was also able to reasonably accurately predict the soil gravimetric water contents (top 75 mm) of sites of contrasting aspect and slope over a four year period. Overall, dryland hill country provides an extremely variable pasture production system which is affected by a great number of factors. The greatest factor affecting pasture growth appears to be plant available N, which in dryland hill country is inherently severely deficient. The other major factors are mostly climate-related and vary between sites depending on aspect and slope location. The intensification of farming systems demands that these variations can be accurately accounted for so that more accurate whole farm system models can be used to develop farming policy and economic models. Calculating E, and the subsequent depression from maximum yield by soil fertility, appears to be a method by which future models may more accurately aid farmers in analysing the efficiency of their system.