Summary. The substitution value, or horizontal comparison between a fertiliser and some high-performing reference fertiliser, is shown to be a powerful and effective means of ranking the performance of various reactive phosphate rock fertilisers relative to a water-soluble phosphate such as a superphosphate. To take account of the variability due to random error in fertiliser field trials, the use of a standardised score, one measuring the relative departure of the substitution value from that expected for the standard fertiliser, is shown to be a suitable statistic for general use. Cluster analyses using these z-scores as input data can lead to appropriate decision making such as, for example, grouping experimental sites into: (i) those which perform as well as water-soluble phosphate, (ii) those which are somewhat worse than water-soluble phosphate, and (iii) those which are much worse than water-soluble phosphate. Substitution values obtained for different reactive phosphate rocks were compared both within and between sites, using appropriate statistical tests of significance. This approach assisted in ranking the performance of the 5 reactive phosphate rocks and the partially acidulated phosphate rock studied in the National Reactive Phosphate Rock Project, and helped predict which ones are likely to be suitable for use in differing environments.