The catalytic decomposition of ammonia has recently been proposed as a possible source of hydrogen for fuel cells. However, the ruthenium catalyst is costly. Although there exist several correlations for catalytic activity that suggest potentially useful alternatives, the particular candidates differ. The present work seeks to determine experimentally which, if any, of these correlations correctly predicts suitable substitutes. The experiments examine 13 different metallic catalysts from numerous places within the Periodic Table, and show that the activity varies in the order Ru>Ni>Rh>Co>Ir>Fe≫Pt>Cr>Pd>Cu≫Te, Se, Pb. The results suggest that nitrogen desorption limits the rate on Fe, Co, and Ni, whereas N–H bond scission limits the rate on other metals such as Rh, Ir, Pd, Pt, and Cu. Conventional single-parameter correlations of activity generally fail to predict the results because the rate-determining step changes across the data set.