Observational monitoring of the stratospheric transport circulation, the Brewer–Dobson circulation (BDC), is crucial to estimate any decadal to long-term changes therein, a prerequisite to interpret trends in stratospheric composition and to constrain the consequential impacts on climate. The transport time along the BDC (i.e. the mean stratospheric age of air, AoA) can best be deduced from trace gas measurements of tracers which increase linearly with time and are chemically passive. The gas sulfur hexafluoride (SF6) is often used to deduce AoA because it has been increasing monotonically since the ∼1950s, and previously its chemical sinks in the mesosphere have been assumed to be negligible for AoA estimates. However, recent studies have shown that the chemical sinks of SF6 are stronger than assumed and become increasingly relevant with rising SF6 concentrations. To adjust biases in AoA that result from the chemical SF6 sinks, we here propose a simple correction scheme for SF6-based AoA estimates accounting for the time-dependent effects of chemical sinks. The correction scheme is based on theoretical considerations with idealized assumptions, resulting in a relation between ideal AoA and apparent AoA which is a function of the tropospheric reference time series of SF6 and of the AoA-dependent effective lifetime of SF6. The correction method is thoroughly tested within a self-consistent data set from a climate model that includes explicit calculation of chemical SF6 sinks. It is shown within the model that the correction successfully reduces biases in SF6-based AoA to less than 5 % for mean ages below 5 years. Tests using only subsampled data for deriving the fit coefficients show that applying the correction scheme even with imperfect knowledge of the sink is far superior to not applying a sink correction. Furthermore, we show that based on currently available measurements, we are not able to constrain the fit parameters of the correction scheme based on observational data alone. However, the model-based correction curve lies within the observational uncertainty, and we thus recommend using the model-derived fit coefficients until more high-quality measurements are able to further constrain the correction scheme. The application of the correction scheme to AoA from satellites and in situ data suggests that it is highly beneficial to reconcile different observational estimates of mean AoA.