An experimental design and a statistical method for the estimation of the clustering-response activity of lymphocytosis-promoting factor (LPF) in Chinese hamster ovary cells growing in wells on a microplate were investigated. The scoring method introduced by Ipsen was adopted to express the grade of the clustering response rather than the end-point method generally used. The scoring method was validated by statistical analyses. The grade of response varied with the location of the wells on a microplate, and thus the expression of the clustering activity of a test sample in terms of the end-point may be inadequate in terms of accuracy and reproducibility. It was shown that the allocation of test samples to individual wells according to a Latin square design minimized the effect of the location of wells on the clustering response. Under such experimental conditions, a fairly precise and reproducible method for the quantification of the clustering activity was developed.