This thesis concerns certain aspects of the production of CCN. In-cloud oxidation followed by cloud droplet evaporation is an important chemical process where sulfate is produced, and only particles with radii bigger than a certain critical radius (rc) can go through this process. The original interval for the critical radius (rcmin = 0.05 ìm, rcmax = 0.2 ìm) has led to an overestimation of the concentration of CCN. The aim is therefore to develop a new parameterization of the critical radius and at the same time keep the consistency with realistic supersaturations in different types of clouds. An interval for the critical radius must be used because of the great variability of supersaturations. Based on the assumption that the critical radius must equal the critical radius for activation of CCN, two sets of intervals has been developed where the first is consistent with a marine background (rcmin = 0.033 ìm, rcmax = 0.060 ìm), and the second is consistent with a continental background (rcmin = 0.044 ìm, rcmax = 0.16 ìm). These results were implemented in NCAR CCM3 (CCM-Oslo), from which results concerning cloud droplet concentration, effective cloud droplet radius an short wave indirect cloud forcing were made for the year of 2001. Comparisons have been made with earlier results where the original interval of the critical radius has been used. Both the marine and the continental values of the critical radius are smaller than the original values, with the marine values being the smallest. This led to a greater global mean of the cloud droplet concentration (123.4 cm-3 for the marine rc-interval), a smaller global mean of the effective cloud droplet radius (10.6 ìm for the marine rc-interval) and a greater short wave cloud forcing (-54.82 Wm-2 for the marine rc-interval). Considerable parts of South-America, Northern and Central parts of Africa and Northern parts of Australia proved to have smaller concentrations of cloud droplets, and hence larger effective droplet radius and smaller short wave cloud forcing, for the smallest choice of rcmin. The biases were small, especially for the cloud droplet concentration, and are probably results of a numerical error in the parameterizations of aerosol properties.