This paper applies hedonic two-stage technique to investigate the effectiveness of the benefit-transfer methodologies in a statistically controlled repeated sampling experiment. Benefit transfers are carried out on several sets of one thousand pairs of random, non-overlapping sub-samples drawn from a Chicago housing sample. The study finds that even under extremely favorable conditions, quality of transferred benefits is rather poor. Moreover, there is no gain in precision in adopting function transfer over unit-value transfer. The quality of transferred benefits deteriorates in the presence of dissimilarities between two sites, and function-transfer appears more appropriate than unit-value transfer under this situation.