Determining the true, kinetically dominant catalytically active species, in the classic benzene hydrogenation system pioneered by Maitlis and co-workers 34 years ago starting with [RhCp*Cl-2](2) (CP* = [eta(5)-C-5(CH3)(5)]), has proven to be one of the most challenging case studies in the quest to distinguish single-metal-based "homogeneous" from polymetallic, "heterogeneous" catalysis. The reason, this study will show, is the previous failure to use the proper combination of: (i) in operando spectroscopy to determine the dominant form(s) of the precatalyst's mass under catalysis (i.e., operating) conditions, and then crucially also (ii) the previous lack of the necessary kinetic studies, catalysis being a "wholly kinetic phenomenon" as J. Halpern long ago noted. An important contribution from this study will be to reveal the power of quantitiative kinetic poisoning experiments for distinguishing single-metal, or in the present case subnanometer Rh-4 cluster-based catalysis, from larger, polymetallic Rh(0)(n) nanoparticle catalysis, at least under favorable conditions. The combined in operando X-ray absorption fine structure (XAFS) spectroscopy and kinetic evidence provide a compelling case for Rh-4-based, with average stoichiometry "Rh4Cp*2.4Cl4Hc", benzene hydrogenation catalysis in 2-propanol with added Et3N and at 100 degrees C and 50 atm initial H-2 pressure. The results also reveal, however, that if even ca. 1.4% of the total soluble Rh(0)(n) had formed nanoparticles, then those Rh(0)(n) nanoparticles would have been able to account for all the observed benzene hydrogenation catalytic rate (using commercial, ca. 2 nm, polyethyleneglycol-dodecylether hydrosol stabilized Rh(0)(n) nanoparticles as a model system). The results-especially the poisoning methodology developed and employed-are of significant, broader interest since determining the nature of the true catalyst continues to be a central, often vexing issue in any and all catalytic reactions. The results are also of fundamental interest in that they add to a growing body of evidence indicating that certain, appropriately ligated, coordinatively unsaturated, subnanometer M-4 transition-metal clusters can be relatively robust catalysts. Also demonstrated herein is that Rh4 clusters are poisoned by Hg(0), demonstrating for the first time that the classic Hg(0) poisoning test of "homogeneous" vs "heterogeneous" catalysts cannot distinguish Rh-4-based subnanometer catalysts from Rh(0)(n) nanoparticle catalysts, at least for the present examples of these two specific, Rh-based catalysts.