Seven lower Pliocene sediment samples from Ocean Drilling Program (ODP) Hole 642B on the outer V0ring Plateau were analyzed in order to test the reproducibility of Moore's random settling method in quantitative diatom analysis. The results indicate that the absolute abundance of diatom valves per gram dry sediment are obtained within an error of ?5% at the 95% confidence level after a count of 400 valves. The results also indicate that relative abundance percentages for abundant species (approximately 20% of the total assemblage) are determined with a maximum error of ?3.5%, whereas relative abundance percentages for less abundant species are determined with a maximum error of ?1.5%. Species having relative abundances of approximately two percent or less may be overlooked. This study demonstrates that absolute abundance data are required in order to understand the cause of change in the relative abundance. INTRODUCTION The usefulness of quantitative methods in diatom biostratigraphy and paleoceanography has been demonstrated by, for example, Burckle and Trainer (1979). They were using relative abundances to define Pliocene diatom datums not only by first and last appearances, but also by recurrences and by changes in dominance of two closely related species (see also Baldauf 1985; Barron 1985). Counts of total diatom abundance changes in marine sediments provide valuable paleoceanographic information. For example, Stabell (1986) analyzed absolute abundances of diatoms in Quaternary sediments for recognition of high productivity events that were caused by movements of the oceanic polar front (see also Pokras and Ruddiman 1989; Stabell 1989). Among described methods that are used to calculate diatom abundances per weight or volume of unit sediment are the evaporation technique (Batterbee 1973), the pipetting technique (Schrader and Gersonde 1978), and the use of marker grains (Kaland and Stabell 1981). The random settling method described by Moore (1973) was developed for studies of radiolarian abundances. Laws (1983) tested the accuracy of the method by using a known mixture of microspheres, and showed that this method produced accurate strewn mounts within the limits of error relative to the original sample. In contrast, pipetted mounts were biased toward larger particles (i.e. larger particles were obtained at frequencies higher than the occurrence within the original sample). Moore's method can also be adopted to estimate absolute abundances of diatoms. The major purpose of this study is to test empirically the reproducibility of Moore's method for use in quantitative diatom analysis, using both sample splitting and settling containers. Another purpose is to evaluate how absolute abundance changes are reflected in a relative abundance analysis. In order to achieve these goals, both absolute and relative abundances were analyzed in seven samples, and counts were performed on multiple slides from each sample. This allows comparison between absolute and relative abundance changes, and the reliability of the obtained values can be evaluated.