The relative importance of viral lysis and bacterivory as causes of bacterial mortality were estimated. A laboratory experiment was carried out to check the kind of control that viruses could exert over the bacterial assemblage in a non-steady-state situation. Virus-like particles (VLP) were determined by using three methods of counting (DAPI [4',6-diamidino-2-phenylindole] staining, YOPRO staining, and transmission electron microscopy). Virus counts increased from the beginning until the end of the experiment. However, different methods produced significantly different results. DAPI-stained VLP yielded the lowest numbers, while YOPRO-stained VLP yielded the highest numbers. Bacteria reached the maximal abundance at 122 h (3 x 10(7) bacteria ml-1), after the peak of chlorophyll a (80 μg liter-1). Phototrophic nanoflagellates followed the same pattern as for chlorophyll a. Heterotrophic nanoflagellates showed oscillations in abundance throughout the experiment. The specific bacterial growth rate increased until 168 h (2.6 day-1). The bacterivory rate reached the maximal value at 96 hours (0.9 day-1). Bacterial mortality due to viral infection was measured by using two approaches: measuring the percentage of visibly infected bacteria (%VIB) and measuring the viral decay rates (VDR), which were estimated with cyanide. The %VIB was always lower than 1% during the experiment. VDR were used to estimate viral production. Viral production increased 1 order of magnitude during the experiment (from 10(6) to 10(7) VLP ml-1 h-1). The percentage of heterotrophic bacterial production consumed by bacterivores was higher than 60% during the first 4 days of the experiment; afterwards, this percentage was lower than 10%. The percentage of heterotrophic bacterial production lysed by viruses as assessed by the VDR reached the highest values at the beginning (100%) and at the end (50%) of the experiment. Comparing both sources of mortality at each stage of the bloom, bacterivory was found to be higher than viral lysis at days 2 and 4, and viral lysis was higher than bacterivory at days 7 and 9. A balance between bacterial losses and bacterial production was calculated for each sampling interval. At intervals of 0 to 2 and 2 to 4 days, viral lysis and bacterivory accounted for all the bacterial losses. At intervals of 4 to 7 and 7 to 9 days, bacterial losses were not balanced by the sources of mortality measured. At these time points, bacterial abundance was about 20 times higher than the expected value if viral lysis and bacterivory had been the only factors causing bacterial mortality. In conclusion, mortality caused by viruses can be more important than bacterivory under non-steady-state conditions.