<p><strong>Abstract.</strong> In the atmospheric boundary layer (ABL), incomplete mixing (i.e., segregation) results in reduced chemical reaction rates compared to those expected from mean values and rate constants derived under well mixed conditions. Recently, segregation has been suggested as a potential cause of discrepancies between modelled and measured OH radical concentrations, especially under high isoprene conditions. Therefore, the influence of segregation on the reaction of OH radicals with isoprene has been investigated by modelling studies and one ground-based and one aircraft campaign.</p> <p>In this study, we measured isoprene and OH radicals with high time resolution in order to directly calculate the influence of segregation in a low-NO<sub>x</sub> and high-isoprene environment in the central Amazon basin. The calculated intensities of segregation (<i>I</i><sub>s</sub>) at the Amazon Tall Tower Observatory (ATTO) above canopy top are in the range of values determined at a temperate deciduous forest (ECHO-campaign) in a high-NO<sub>x</sub> low-isoprene environment, but stay below 10&#8201;%. To establish a more general idea about the causes of segregation and their potential limits, further analysis was based on the budget equations of isoprene mixing ratios, the variance of mixing ratios, and the balance of the intensity of segregation itself. Furthermore, it was investigated if a relation of <i>I</i><sub>s</sub> to the turbulent isoprene surface flux can be established theoretically and empirically, as proposed previously. A direct relation is not given and the amount of variance in <i>I</i><sub>s</sub> explained by the isoprene flux will be higher the less the influence from other processes (e.g., vertical advection) is and will therefore be greater near the surface. Although ground based values of <i>I</i><sub>s</sub> from ATTO and ECHO are in the same range, we could identify different dominating processes driving <i>I</i><sub>s</sub>. For ECHO the normalized variance of isoprene had the largest contribution, whereas for ATTO the different transport terms expressed as a residual were dominating. To get a more general picture of <i>I</i><sub>s</sub> and its potential limits in the ABL, we also compared these ground based measurements to ABL modelling studies and results from an aircraft campaign. The ground based measurements show the lowest values of the degree of inhomogenous mixing (<&#8201;20&#8201;%, mostly below 10&#8201;%). These values increase if the contribution of lower frequencies is added. Values integrated over the whole boundary layer (modelling studies) are in the range from 10&#8201;% to 30&#8201;% and aircraft measurements integrating over different landscapes are amongst the largest reported. This presents evidence that larger scale heterogeneities in land surface properties contribute substantially to <i>I</i><sub>s</sub>.</p>