Different schemes have been used to represent the vertical coherence of clouds for radiation and cloud microphysics in general circulation models and for photolysis in photochemical transport models. Here, we examine the maximum-random overlap scheme, the random overlap scheme, and a linear scheme (linear scaling of the cloud optical depth in a grid box with cloud fraction) and evaluate their effects on averaged photolysis frequencies and OH concentrations in a global photochemical model. Photolysis frequencies are increased in the upper tropical troposphere and decreased in the lower troposphere if clouds are assumed to be randomly overlapped or if a linear assumption is followed rather than the maximum-random overlap assumption. The underestimate is of order 50% and 20--30% at the surface in the tropics and over some continental midlatitude regions for the linear and the random assumptions, respectively, relative to the maximum-random assumption. The global average CH4 + OH reaction rate-weighted OH concentration calculated with the random overlap assumption is within a few percent of that calculated with the maximum-random overlap assumption but is underestimated with the nonphysically based linear assumption, by 4 to 6% relative to the maximum-random overlap assumption and by 6 to 7% relative to the random overlap assumption. An underestimate of rate-weighted OH concentration implies that the tropospheric lifetime for CH4 and other greenhouse gases that react with OH is too long in models that use this scheme.