The 17O excess (17Δ) of dissolved O2 has been used, for over a decade, to estimate gross O2 production (G17OP) rates in the mixed layer (ML) in many regions of the ocean. This estimate relies on a steady-state balance of O2 fluxes, which include air-sea gas exchange, photosynthesis and respiration but notably, not turbulent mixing with O2 from the thermocline. In light of recent publications, which showed that neglecting the turbulent flux of O2 from the thermocline may lead to inaccurate G17OP estimations, we present a simple correction for the effect of this flux on ML G17OP. The correction is based on a turbulent-flux term between the thermocline and the ML, and use the difference between the ML 17Δ and that of a single data-point below the ML base. Using a numerical model and measured data we compared turbulence-corrected G17OP rates to those calculated without it, and tested the sensitivity of the GOP correction for turbulent flux of O2 from the thermocline to several parameters. The main source of uncertainty on the correction is the eddy-diffusivity coefficient, which induces an uncertainty of ~50%. The corrected G17OP rates were 10-90% lower than the previously published uncorrected rates, which implies that a large fraction of the photosynthetic O2 in the ML is actually produced in the thermocline.