Total direct counts of bacterial abundance are central in assessing the biomass and bacteriological quality of water in ecological and industrial applications. Several factors have been identified that contribute to the variability in bacterial abundance counts when using fluorescent microscopy, the most significant of which is retaining an adequate number of cells per filter to ensure an acceptable level of statistical confidence in the resulting data. Previous studies that have assessed the components of total-direct-count methods that contribute to this variance have attempted to maintain a bacterial cell abundance value per filter of approximately 106 cells filter-1. In this study we have established the lower limit for the number of bacterial cells per filter at which the statistical reliability of the abundance estimate is no longer acceptable. Our results indicate that when the numbers of bacterial cells per filter were progressively reduced below 105, the microscopic methods increasingly overestimated the true bacterial abundance (range, 15.0 to 99.3%). The solid-phase cytometer only slightly overestimated the true bacterial abundances and was more consistent over the same range of bacterial abundances per filter (range, 8.9 to 12.5%). The solid-phase cytometer method for conducting total direct counts of bacteria was less biased and performed significantly better than any of the microscope methods. It was also found that microscopic count data from counting 5 fields on three separate filters were statistically equivalent to data from counting 20 fields on a single filter.