In this paper we study the effect of turbulence on the oxidation rate of isoprene and its reaction products in the atmospheric boundary layer. We use two models of different complexity: a simple model consisting of two well-mixed layers and a one-dimensional off-line second-order closure model. Both models include an extensive set of chemical reactions to describe the oxidation of isoprene. A 5-day simulation is performed to compare the simple model output with data from the Amazon Boundary Layer Experiment (ABLE-2A). The model is able to represent fairly the basic dynamics and chemistry during this experiment. Subsequently, the simple model provides boundary and initial conditions for a one-dimensional second-order closure model that is used to assess the impact of higher-order chemistry terms on turbulent mixing and chemical transformations. We focus on covariances of NO with (peroxy-) radicals and covariances of OH with stable intermediate products. We find only small effects on the effective reaction rates due to the OH covariances. A significant effect is found of the covariances of NO, inhibiting the effective reaction rates with the peroxy radicals by a maximum of 10% in the afternoon. The inclusion of covariance terms resulted in an increase of radical concentrations, but the NO concentration profiles remained unchanged. Higher-order chemistry terms do have an effect on NO and NO2 fluxes, which change by 5 to 30% in the middle of the boundary layer. Therefore these terms have to be taken into account when flux-gradient relationships or deposition velocities are derived from observations. The present results indicate that the incorporation of higher-order chemistry terms is not essential for a correct representation of the mean profiles of most stable species involved. ¿ 2000 American Geophysical Union