The 40-year reanalysis data set ERA-40 from the European Centre for Medium-Range Weather Forecasts includes, unlike ERA-15, archived convective mass fluxes. These convective fluxes are useful for off-line chemistry transport modeling. The impact of using these archived convective mass fluxes (based on a convective parameterization described in Gregory et al. <2000>) instead of off-line diagnosed mass fluxes (based on a convective parameterization described in Tiedtke <1989>) was investigated with the chemistry transport model TM3. At first sight the two types of mass fluxes look similar. However, some differences can be noted: the archived updrafts extend higher than the off-line diagnosed ones; they are also less intense below 500 hPa over sea. The archived downdrafts are much weaker than the off-line diagnosed downdrafts. With archived convective mass fluxes, we found slightly higher 222Rn concentrations in the boundary layer, lower 222Rn values in the free troposphere and significantly higher 222Rn values in the upper troposphere and lower stratosphere. The effect on tropospheric chemistry of using archived mass fluxes instead of diagnosed ones is an increase of NOx and O3 in the free troposphere, but a decrease in the upper troposphere. The differences amount to up to 20% for O3 in the zonal and seasonal mean. Our results thus underline the sensitivity of tropospheric ozone chemistry to the description of convective transport. Comparison with 222Rn observations shows that the archived convective mass fluxes give better agreement in the tropical upper troposphere. More comparisons to free tropospheric observations of 222Rn or another tracer of convective transport will be needed to unambiguously identify either of the convective data sets as optimal for use in chemistry transport models.