Large scale, quantitative information about the variability of target soil properties is required for forest management. This project is attempting to determine whether or not the New Zealand Soil Classification system (NZSC), when used in combination with a soillandscape model, adequately communicates this information. In the first phase of this project a soil-landscape model was developed and a pilot variability study conducted. The soils in the study area, located in the W oodlaw Block of the Longwood Range, are formed from either Permian andesite or greywacke on moderately steep to steep hill slopes under a moist cool climate and a vegetation cover of beech and podocarp forests. The soil-landscape model was developed using the land systems approach. The model consists of predictive relationships between topographic features and soil classes. There is a clear relationship between slope steepness, the abundance of surface boulders and the gravel content of the soil. A soil-landscape unit map showing the distribution of predicted soil classes has been produced. The results of the pilot variability study have showed that the soils sampled are acidic and have moderate to high P-retention values. An analysis of variance indicated that both of these properties are significantly variable between sites and between horizons. There appears to be a relationship between land component type and the magnitude and variability of these properties. The clay mineralogical analysis revealed that the dominant clay minerals present in all the soils sampled are chlorite-vermiculite, kaolinite, sepiolite, and allophane. The presence of allophane and kaolinite may be related to the moderate to high P-retention values.