Microscopic estimation of bacterial biomass requires determination of both biovolume and biovolume-to-biomass conversion. Both steps have uncertainty when applied to the very small bacteria typically found in natural seawater. In the present study, natural bacterioplankton assemblages were freshly collected, passed through 0.6-mum-pore-size Nuclepore filters to remove larger particulate materials, and diluted for growth in 0.22-mum-pore-size Millipore filter-sterilized unenriched seawater. This provided cells comparable in size and morphology to those in natural seawater, but the cultures were free of the interfering particulate detritus naturally present. Cells were collected on glass-fiber GF/F filters, and biovolumes were corrected for cells passing these filters; C and N were measured with a CHN analyzer. Our criteria for size measurement by epifluorescence photomicrography were confirmed with fluorescent microspheres of known diameters. Surprisingly, in six cultures with average per-cell biovolumes ranging from 0.036 to 0.073 mum, the average per-cell carbon biomass was relatively constant at 20 +/- 0.08 fg of C (mean +/- standard error of the mean). The biovolume-to-biomass conversion factor averaged 0.38 +/- 0.05 g of C cm, which is about three times higher than the value previously estimated from Escherichia coli, and decreased with increasing cell volume. The C:N ratio was 3.7 +/- 0.2. We conclude that natural marine bacterial biomass and production may be higher than was previously thought and that variations in bacterial size may not reflect variations in biomass per cell.