Recent studies of precipitation, aircraft icing, and visibility through fog have focussed attention on the physical constitution of clouds, a subject to which knowledge of the drop-size spectrum and its origin would be an important contribution. The drop-size spectrum resulting when air containing condensation nuclei is uniformly cooled may be computed, leading to a differential equation for the growth of a cloud drop which cannot be integrated analytically. A numerical method of integration is therefore employed. Three integrations of drop growth under simulated natural conditions are described and compared with data for natural clouds, leading to the following conclusions: The computed drop-size spectra agree with observation regarding the relatively uniform clouds most frequently observed. Convergence and mixing of fine-grained turbulence are suggested as influences broadening the distribution in the less uniform clouds. Even for the latter, the computed spectra agree with observation on the shape of the spectrum curve. It is not clear what conditions of cooling most favor a broad spectrum. The computed mean drop sizes agree very well with observation, indicating that growth by collision is not ordinarily significant in uniform clouds. The concentration of drops is determined primarily by the rate of cooling during the initial stage of condensation. It depends, as a rule, only slightly on the concentration of condensation nuclei. With continued uniform cooling, the drop concentration diminishes slightly toward a fixed constant value. It can be increased, once the cloud has formed, only through a great increase in the rate of cooling. Supersaturation during cloud formation ordinarily reaches about 0.1 per cent and can surpass 1 per cent only under extreme circumstances. Subsaturations in descending air currents have the same order of magnitude. Observations of lower relative humidities in clouds, if real, must indicate clear spaces. Operation of the equation of growth on nuclei of modal sizes proposed by Dessens cannot explain the modal drop sizes reported by Köhler.