In experiments designed primarily to investigate viral lysis, we found that the presence of viruses had a positive effect on the growth of Synechococcus. A Landry-Hassett-type stepwise dilution experiment conducted during a Synechococcus bloom in the Gulf of Mexico used both (i) 0.2-mu m filtered seawater in which the abundance of bacteria and grazers were reduced but the majority of viruses were retained, and (ii) ultrafiltered (30 000 MW cutoff) virus-free seawater in which the abundance of viruses, bacteria and grazers were reduced. High growth rates and frequency of dividing cells (FDCs) were recorded in 0.2-mm filtered treatments while growth was inhibited in incubations with a high proportion of virus-free ultrafiltered water. In two subsequent experiments using Mediterranean Sea populations, a two-point dilution approach in which viral abundance was reduced by 80-90% yielded similar results, and showed that Synechococcus only grew well in the presence of viruses, bacteria and grazers. In four further Mediterranean experiments viruses removed via ultrafiltration were added back, either untreated, or inactivated by a heat treatment. Growth rates and FDCs were higher in the presence of untreated viruses than with viruses inactivated by heat, suggesting that it was not organic matter in the virus-size fraction but rather the presence of infectious viruses which sustained growth. While Synechococcus was also infected by viruses during these experiments, our data imply that growth of Synechococcus may depend upon viral lysis of heterotrophic bacteria. This finding is consistent with the view that nutrient cycling by viral lysis of heterotrophic bacteria may control phytoplankton growth and ecosystem scale carbon production.