In the semi-arid wheat production areas of the Free State province, the low and erratic rainfall, especially during early spring, is the most limiting natural resource for rainfed wheat production. Accurate simulation of soil water content by crop growth simulation models is therefore critically important. Validation data were collected on four benchmark ecot9pes for rainfed wheat production in the Free State province of South Africa over a three-year period. The trials were planted in a randomized block design with four replications. Three cultivars were planted and tested on each ecotope. Plant components, yield components and soil water content measurements were done according to IBSNAT standards for a minimum data set for model validation. Soil water content was monitored continuously with a neutron water meter during the growing season, with more emphasis on the warmer and wetter part of it. The soil physical and chemical properties required as input by the model were determined. Measured data were compared with simulated data, as produced by the standard version of CERES-wheat version 2.10, using graphical and statistical methods. Simulation of the soil water content through the growing season was generally reasonably accurate, but over and under-predictions occurred during critical growth stages. The simulation of plant and yield components was not accurate and large over and under-predictions occurred.In die Vrystaat is die voorkoms en tydigheid van lentereens en die wisselvallige verspreiding daarvan deur die groeiseisoen die mees beperkende faktor in droelandkoringproduksie. Om die rede word uitsonderlik baie op opgegaarde grondwater voor planttyd staatgemaak om die kans op 'n ekonomiese opbrengs te verseker. Die akkurate simulasie van grondwaterinhoud deur gewasgroeisimulasiemodelle is dus krities belangrik om die groei en ontwikkeling van die koringplant in die Vrystaat akkuraat te simuleer. Validasiedata is op vier verteenwoordigende ekotope in die Vrystaat oor 'n tydperk van drie jaar ingesamel. Die proewe is in 'n ewekansige blokontwerp met vier herhalings geplant. Drie kultivars is in elke proef geplant. Metings is uitgevoer volgens IBSNAT-voorskrifte vir die insameling van 'n minimum datastel vir die toets van simulasiemodelle. Grondwaterinhoud is periodiek gedurende die groeiseisoen gemeet. Gemete data is met die simulasiedata van CERES-wheat 2.10 vergelyk deur gebruik te maak van grafiese en statistiese metodes. Die simulasie van grondwaterinhoud is oor die algemeen" redelik akkuraat, maar oor-en ondervoorspellings kom voor tydens kritieke groeistadia. Die simulasie van plant-en oeskomponente is onakkuraat, met groot oor-en ondervoorspellings wat voorkom."