This paper explores a new approach to estimating atmospheric hydroxyl radical concentrations from regional measurements of a suite of hydrocarbons. The approach is guided by the study of a suite of synthetic tracers, i, with uniform continental sources and constant but different lifetimes of 1, 2, 5, 20, 50, and 100 days, whose global distributions are calculated from a three-dimensional chemical tracer model. With the help of the model we show that in a grid box the standard deviation &sgr;i divided by the average concentration Mi is a unique function of the chemical lifetime &tgr;i. In favorable cases, for instance, in surface air within a specific region sampled by the Pacific Exploratory Mission (PEM) West B campaign, that function takes a simple form: &sgr;i/Mi=A¿&tgr;i-α, with α=0.48, very close to 1/2. An analogous relation is found for the alkanes, ethane through n-hexane, measured during the PEM West B campaign in the same domain, with their reaction rate constant with OH, kOH. That relation has the form &sgr;i/Mi=B¿(kOH,i)α', with α'=0.49. Using the alkenes, for example propene, which also react with O3, the dependence on kOH,i can be related to a dependence on &tgr;i. This allows us to estimate the OH concentration, 6¿105 cm-3, with an error of roughly a factor of 2 for this region (boundary layer, 30 ¿N--40 ¿N latitude, and 135 ¿E--155 ¿E longitude in March). This estimate is essentially based on empirical relations only and the assumption that the considered hydrocarbons have the same source distribution. As a by-product, we show that the α defined above is related to the slope in the (logarithmic) correlation plot between the mixing ratios of two trace gases with different lifetimes. We also show that the global distribution of α appears to be a useful tool to diagnose fast regional transport. ¿ 1998 American Geophysical Union