Ship tracks are long-lived, linear regions of enhanced reflectivity in low-lying marine clouds that appear in satellite imagery downwind of ships. Ship tracks were first observed as cloud lines in visible satellite imagery (type 1). A second (and more common) type of ship track (type 2), which is masked at visible wavelengths by natural variability in cloud reflectivity, is seen at near-infrared wavelengths in satellite imagery. A one-dimensional numerical model is used to simulate measurements of both types of ship tracks and to investigate interactions between aerosol and cloud microphysics, radiative transfer, and turbulent mixing in the cloud-topped marine boundary layer that lead to the formation and provide for the persistence of ship tracks. We find that cloud condensation nuclei (CCN) injections can account for many of the observed properties of ship tracks. Higher CCN concentrations produce increased droplet concentrations, which enhance cloud reflectivity by reducing droplet radius and increasing droplet cross-sectional area. The smaller droplets also reduce the drizzle rate, which can allow cloud water to increase under some conditions, thereby leading to higher cloud reflectivity. However, smaller droplets also evaporate more readily below cloud base. Increased evaporation reduces mixing between the cloud and the subcloud layers during daytime, which causes a decrease in cloud water. The distinction between the two types of ship tracks is suggested to be due to differences in ambient concentrations of CCN that cause variations in turbulent mixing in the boundary layer, through the effect of cloud droplet concentrations on cloud-top longwave radiative cooling. The model predicts lifetimes of >1 day and >2 days for the simulated type 1 and type 2 ship tracks, respectively. In the atmosphere, processes not treated in the model, such as horizontal dispersion and changes in large-scale atmospheric conditions, may limit ship track lifetimes. ¿ American Geophysical Union 1995