This paper deals with the ultimate limits of chemical contributions to flame inhibition. Particular attention is focussed on the inhibition cycles which regenerate the inhibitor. This leads to the definition of an idealized "perfect" inhibition cycle. It is demonstrated that for such an inhibitor in a stoichiometric methane/air flame, additive levels in the 0.001-0.01 mole percent range will lead to a decrease in flame velocity of approximately 30%. This efficiency corresponds roughly to the observed behavior of metallic inhibitors such as iron pentacarbonyl which is known to be as much as 2 orders of magnitude more effective than currently used suppressants. This correspondence between the behavior of a "perfect inhibitor" and iron carbonyl leads to the conclusion that only gas-phase processes can account for its inhibitive power.