The role of heterotrophic nanoflagellates (HNAN, size 2-20μm) in grazing on bacteria and mineralization of organic matter in stratified Lake Vechten was studied.Quantitative effects of manipulation and fixation on HNAN were checked. Considerable losses were caused by centrifugation, even at low speed, and by filtration if the vacuum exceeded 3 kPa. HNAN were well preserved for several weeks with unbuffered 5% formaldehyde as well as with 0.3 to 1% glutaraldehyde (final concentrations). However, the chlorophyll autofluorescence of phototrophic nanoflagellates was strongly decreased by formaldehyde, whereas it could be preserved for a few days with 1% glutaraldehyde. Loss of autofluorescence and of cells was prevented for at least one year if prepared primulin-stained slides were stored in a freezer.During summer stratification in 1985, HNAN showed peak densities in the microaerobic lower metalimnion at 6-7 m depth. In this stratum, high HNAN numbers (more than 30 x 10 6HNAN.l -1) coincided with high bacterial production, as estimated by thymidine incorporation. Despite the high production, bacterial numbers (about 5 x 10 9bacteria.l -1) did not peak in the metalimnion but were similar throughout the whole water column, indicating that the produced bacteria were consumed by HNAN. A steep decline of HNAN was followed by a peak of ciliates. Thus, HNAN seemed to be a link in the food web between bacteria and ciliates. Batch incubations with selectively filtered lake water, with bacteria alone- and with bacteria plus protozoa, indicated ingestion rates of about 7 bacteria.HNAN -1.h -1, and suggested that the HNAN are able to balance the estimated bacterial population doubling times of about 1 day.The breakdown of organic carbon (C) of dead cyanobacterial cell walls added to selectively filtered lake water was not accelerated by protozoa. However, HNAN and ciliates strongly increased the mineralization of nitrogen (N) and especially phosphorus (P), both from Synechococcus cell walls and cell extract. Bacterial growth did not result in net P mineralization, but in phosphate uptake. P was remineralized only in the presence of protozoan grazers. Grazers increased N mineralization too, although N was also mineralized by bacteria in the absence of grazers. Nuclepore filters were found to release dissolved organic carbon during selective filtration.In two-stage continuous cultures, ingestion rates of HNAN increased from 2.3 to 9.2 bacteria.HNAN -1.h -1, at specific growth rates from 0.15 to 0.65 day -1.On a yeast extract medium with a molar C/N/P ratio of 100/15/1.2 (Redfield ratio), a mixed bacterial population had a gross growth efficiency of 18% (C/C) and a specific carbon content of 211 fg C .μ m -3, at a biovolume of 0.14 C .μ m -3. The HNAN carbon content, biovolume and yield were estimated to be 127 C .μ m -3, 40 C .μ m -3and 47% (C/C). Although p was not growth limiting, HNAN accelerated the mineralization of P from dissolved organic matter 6-fold. The major mechanism of P remineralization appeared to be direct consumption of bacteria by HNAN. N mineralization was performed mainly (70%) by bacteria, but was increased 30% by HNAN. HNAN did not enhance the decomposition of the relatively mineral- rich dissolved organic matter. An accelerated decomposition of organic carbon by protozoa may be restricted to mineral-poor substrates, and may be explained mainly by protozoan nutrient regeneration. With subsamples from the continuous cultures, protozoan grazing was found to be accurately measured by uptake of fluorescently labelled bacteria.In continuous culture an empirical relationship between bacterial cell production and thymidine incorporation was established. Tritium-labelled thymidine incorporation into total cold-TCA-insoluble macromolecules yielded a relatively constant empirical conversion factor of about 1 (range 0.38-1.42) x 10 18bacteria produced per mol of thymidine incorporated, at specific growth rates ranging from 0.007 to 0.116 h -1. Although thymidine incorporation has been assumed to be a measure of DNA synthesis, thymidine incorporation appeared to underestimate the independently measured DNA synthesis at least by 1.5 to 13-fold, even if all incorporated label was assumed to be in DNA. However, incorporation into DNA was found to be insignificant, as measured by conventional acid-base hydrolysis. These observations suggest that the empirical relationship between thymidine incorporation and bacterial growth rate is not based primarily on DNA synthesis.In 1987, thymidine incorporation into total macromolecules yielded maximum bacterial production rates of 141 x 10 6bacteria.l -1.h -1(4.2μg C.l -1.h -1) in the metalimnion, which was one order of magnitude higher than in the epi- and hypolimnion. In all three strata, the estimated bacterial production was roughly balanced by the in-situ protozoan grazing determined with fluorescently labelled bacteria. HNAN showed ingestion rates of 2 to 17 bacteria.HNAN -1.h -1. and were the main consumers of bacteria.The moderate bacterial densities, high bacterial growth rates, high HNAN densities and high grazing rates in the metalimnion are indicative of protozoan control of bacterial numbers. The bacterial secondary production in the metalimnion is estimated to be about 20% of the phytoplankton primary production, and may be largely supported by extracellular release of labile photosynthate and decaying plankton.