We present the results of a 1D modelling study using a two equation k-epsilon turbulence closure model to generate realistic upper-ocean temperatures. The model is forced hourly with a variety of idealized diurnal wind and cloud profiles to investigate the sensitivity of the diurnal response and its effect on air-sea CO2 transfer at low to moderate wind speeds. In many cases a diurnal warm layer is formed and this affects gas fluxes. The results show that phase-dependent wind speed and cloud fraction with respect to solar insolation are important for determining the magnitude and shape of the diurnal variability and the depth of the diurnal mixed layer. Phasing the wind forcing could double the diurnal temperature difference (e.g. for solar insolation 12 h out of phase with maximum wind stress), and in the same case the net CO2 flux was increased by approximately 30%. Exact calculations of air-sea CO2 flux are compared to simplified calculations based on night-time temperatures, with or without a 'cool-skin' correction and average wind speed. These simplified calculations underestimate the net CO2 flux in a weak source region. When AT was at it largest the net daily air-sea CO2 flux was reduced by up to 61.7% if daily average values for wind speed and SST were used, compared to the flux obtained from the I D model. Shortterm wind speed variability leads to large discrepancies as a result of the non-linear relationship between wind speed and CO2 transfer, which is even further enhanced by diurnal processes and subsequent convective enhancement. Irrespective of varying winds or clouds, the formation of a diurnal warm layer has a significant effect on air-sea CO2 transfer resulting in increased ocean degassing. Crown Copyright (C) 2008 Published by Elsevier Ltd. All rights reserved.