This study compares the efficiency of pot trials in the glasshouse or outdoors, and trials on undisturbed soil cores (100 cm deep; 15 cm wide), at simulating field trials. Yields of spring barley (Hordeum vulgate, cv Julia) were measured at sites on seven soil series on 11 farms, and also on undisturbed soil cores from the same sites, and in pot trials using disturbed soil from the same sites, to ascertain how far data from the core and pot trials (simulation procedures) could be used to generate the models, regressions or generalisations that were produced by more laborious and time-consuming field trials. There was no evidence of interactions between the soil series and simulation procedures. Correlations between field yields and yields from simulation procedures showed that the undisturbed cores gave the highest correlation (r = 0.59**) with field yields. This was followed by yields measured in outdoor pot trials (r = 0.36** and 0.67** in two years), and by pot trials in the glasshouse (r = 0.11 and 0.25, neither significant). In regressions of field yield on yields from simulation trials, the intercept on the field yield axis was lowest for cores and highest for pot trials in the glasshouse. Anovar studies on the partitioning of the total variability in trial yields showed that undisturbed cores gave a partitioning of yields more similar to that of field trials than did the pot trials. In some respects, but not all, core trials also gave the most similar regressions of yield on soil properties to those obtained from field trials.The use of undisturbed cores, collected at sites of interest and carried to a central site, combines the advantages of field trials (natural soil structure and environment) with those of pot trials (standard environment, easier management).