The conditional sampling formulation typically used in eddy accumulation flux measurements was tested at two sites using velocity, temperature, and specific humidity time series measurements. The first site was at Owen's Lake in southern California, and the second site was an irrigated bare soil field at the Campbell Tract facility in Davis, California. The constant &bgr; relating the turbulent flux to the accumulated concentration difference between updrafts and downdrafts was related to the statistics of the vertical velocity. It was found that this formulation for &bgr; reproduces the extreme fluxes (both sensible and latent heat) better than previous &bgr; formulations. The overall comparison between eddy correlation measured fluxes and conditional sampling predicted fluxes for a wide range of atmospheric stability conditions was excellent (R2=0.97). On average, the mean &bgr; obtained from our experiments for temperature and water vapor is in good agreement (¿4%) with other reported &bgr; values for carbon dioxide, water vapor, ozone, and temperature. These results suggest that the conditional sampling formulation used in eddy accumulation methods is useful for routine flux measurements of passive greenhouse gases. ¿ American Geophysical Union 1994