This study analyzes 27 cumuliform and stratiform clouds sampled aboard the CIRPAS Twin Otter during the 2004 International Consortium for Atmospheric Research on Transport and Transformation (ICARTT) experiment. The data set was used to assess cloud droplet closure using (1) a detailed adiabatic cloud parcel model and (2) a state-of-the-art cloud droplet activation parameterization. A unique feature of the data set is the sampling of highly polluted clouds within the vicinity of power plant plumes. Remarkable closure was achieved (much less than the 20% measurement uncertainty) for both parcel model and parameterization. The highly variable aerosol did not complicate the cloud droplet closure, since the clouds had low maximum supersaturation and were not sensitive to aerosol variations (which took place at small particle sizes). The error in predicted cloud droplet concentration was mostly sensitive to updraft velocity. Optimal closure is obtained if the water vapor uptake coefficient is equal to 0.06, but can range between 0.03 and 1.0. The sensitivity of cloud droplet prediction error to changes in the uptake coefficient, organic solubility and surface tension depression suggest that organics exhibit limited solubility. These findings can serve as much needed constraints in modeling of aerosol-cloud interactions in the North America; future in situ studies will determine the robustness of our findings. Copyright 2007 by the American Geophysical Union.