A series of experiments were performed to test the hypothesis that bacterivorous protists selectively feed on specific genotypes of marine bacterioplankton, affecting bacterioplankton community structure. A study comparing the bacterioplankton community diversity estimated by a SSU rDNA gene (SSU rDNA) clone library and a collection of cultivated bacteria from the same water sample suggests that bacterioplankton is dominated by organisms that do not grow easily in enrichment cultures and are underrepresented in culture collections. Therefore, a new method, length heterogeneity analysis by PCR (LH-PCR) was developed to assess the community structure of in situ bacterioplankton communities. In LH-PCR, a region of the SSU rDNA which exhibits length variations among different phylogenetic groups is amplified from environmental DNA by PCR. Fragments originating from different organisms are discriminated by their length and quantified by the fluorescence emission of a labeled primer. Since the method is based on PCR, a study was performed to evaluate the introduction of bias by the reaction. Using pairwise mixtures of rDNAs I described reannealing bias, a new source of PCR bias by PCR. This bias is caused by self-inhibition of PCR amplicons that attain elevated concentrations after several reaction cycles; thus templates with higher concentration in original gene mixtures tend to be underrepresented in products. This bias can be minimized by limiting final PCR product concentrations. I applied LH-PCR to assess changes in bacterioplankton communities in four protist exclusion experiments. Changes in the genotypic composition of the bacterial community of water samples with protists removed by filtration was followed and compared to the changes in community structure of control water samples. There were differences between filtered water samples and controls incubated for 24 to 48 hours. In the absence of bacterivores some SSU rDNAs that were insignificant in the original water samples dominated the bacterioplankton SSU rDNA pool after 48 hours. Protists appeared to be capable of controlling bacterioplankton taxonomic diversity under manipulated conditions. This supports the hypothesis that aquatic bacterioplankton communities are dominated by relatively inactive cells that are less susceptible to grazing by bacterivorous protists.