Bacterial viruses, or bacteriophages, have numerous roles in marine systems. Although they are now considered important agents of mortality of bacteria, a second possible role of regulating bacterial community composition is less well known. The effect on community composition derives from the presumed species-specificity and density-dependence of infection. Although models have described the "kill the winner" hypothesis of such control, there are few observational or experimental demonstrations of this effect in complex natural communities. We report here on some experiments that demonstrate that viruses can influence community composition in natural marine communities. Although the effect is subtle over the time frame suitable for field experiments (days), the cumulative effect over months or years would be substantial. Other virus roles, such as in genetic exchange or microbial evolution, have the potential to be extremely important, but we know very little about them.