A particle-size (psd) distribution is one of the most important attributes in determining soil physical and chemical processes and is increasingly being used with pedotransfer functions to predict soil hydraulic properties. The need for high resolution prediction of psd’s is increasing as models become more distributed and work on a finer scale. The use of ancillary data to increase the prediction precision is becoming more common as remotely sensed data become more available. Conventional methods of prediction however such as cokriging and multiple linear regression do not lend themselves to use with ancillary data and don’t consider the special requirements of a regionalised composition (a collection of variables that must sum to unity or 100%, e.g., psd) of (i) being non negative (ii) summing to a constant at each location and (iii) be an unbiased estimation, and may therefore lead to a false interpretation. The research investigates the use of ancillary data (Radiometric, Electromagnetic and Landsat TM and digital elevation data) to predict soil psd’s using both multiple linear regression (MLR) and generalised additive models (GAM) in a standard from and with an Additive Logratio Transformation (ALR) applied. The results show that the use of ancillary data improved prediction precision by approximately 20% for clay and sand and between 10-14% for silt for all methods when compared to ordinary kriging. However the Additive Log Ratio Transformation method adheres to the special requirements of a composition, with all predicted values non negative and psd’s summing to unity at each prediction point and giving a more accurate textural prediction.