Aircraft-based vertical flux measurements fill a gap in the spatial domain for studies of biosphere–atmosphere exchange. To acquire valid flux data, a determination of the deviation from the mean vertical wind, w', is essential. When using aircraft platforms, flux measurements are subject to systematic and random errors from airflow distortion caused by the lift-induced upwash ahead of the aircraft. Although upwash is typically considered to be a constant quantity over periods used for calculating fluxes, it can vary significantly over short (and longer) periods due to changes in aircraft lift. The characterization of such variations in upwash are of undeniable importance to flux measurements, especially when real-time computations of w' are required. In this paper, the variability in upwash was compared to the calculated upwash from the model of Crawford et al. (Boundary-Layer Meteorol, 80:79–94, 1996) using data taken during a long-period (phugoid mode) free oscillation of the aircraft. The cyclic variation of lift during the free oscillation offers an ideal scenario in which to acquire in-flight data on the upwash that is present, as well as to test the capability of upwash correction models. Our results indicate that while this model corrects for much of the mean upwash, there can be significant variations in upwash on a time scale that is important to flux measurements. Our results suggest that use of the measured load factor could be an easily implemented operational constraint to minimize uncertainty in w' due to changing upwash from changing aircraft lift. We estimate, using the phugoid data, and from variations in aircraft attitude and airspeed in flux-measurement configuration, that the uncertainty in w caused by variable upwash is approximately ± 0.05 m s-1.