We used an analysis of radical propagation efficiency and OH chain length in a simple trajectory model to propose combinations of long-lived species that distinguish conditions in which O3 concentration (<O3>) is NOx-limited and radical-limited. We further examined these indicators in a three-dimensional grid model. We proposed several new indicators including 2O2>/(3>+2>), <O3>/<NOx>, and a measure using the OH rate constant weighted concentrations of NO2 and hydrocarbons. Our analysis also supports the use of several indicators previously proposed by other researchers, including 3>/3> and 2O2>/3>. We found that /2> was more useful than the previously proposed <HCHO>/<NOx>. We found that the indicators easily distinguished extremely NOx-limited or extremely radical-limited regimes but did not reliably distinguish conditions closer to the transition between these two regimes. We propose that a combined analysis using photochemical model simulations and a large set of indicators of both <O3> sensitivity and local odd oxygen production (P(Ox)) sensitivity to VOC and NOx provides the most complete and useful description of <O3> sensitivity. Time series of the indicators, at least from mid-morning to late afternoon, provide useful information about the evolution of <O3> sensitivity during the day. Values of the indicators change depending on the <O3> level due in part to the effects of miscellaneous OH and HO2 termination reactions and to the effects of the composition of the HC mixture on P(Ox). Further evaluation of these indicators using modeling studies, measurements, and test cases with NOx or VOC emissions changes are needed to determine how reliably they distinguish NOx- and radical-limited conditions. ¿ 2000 American Geophysical Union