A series of optimization experiments was carried out to improve visualization of DAPI-stained bacteria extracted from sandy sediments and midgut sediments of the deposit feeder Abarenicola pacifica. Treatments tested for the effects of fixative removal (prior to sonication) and removal of unbound DAPI (prior to filtration). Both quantitative (mean number of cells) and qualitative (cell fluorescence, background fluorescence, dispersion, cell counting rates) characteristics were examined. Removal of fixative and/or unbound stain did not significantly increase mean cell counts versus controls for sandy sediments, although variability (unbiased coefficient of variation) was reduced. Qualitative characteristics were not significantly improved by removing fixative, although visualization improved slightly and counting rates (cells s-1) increased with final removal of stain in midgut sediments, removal of fixative and unbound stain resulted in increased cell counts as well as decreased variability. Significant improvements in contrast (i.e. increased cell fluorescence, decreased background fluorescence) were observed, as well as increased counting rate. These results indicate that both qualitative and quantitative aspects of direct counts may be improved by these methods for problematic sediments such as deposit-feeder gut and fecal samples or highly organic muds.