This study aims to reveal the effects of differential dissolution on the quantitative composition of foraminiferal
assemblages. Through this, we develop a procedure to evaluate dissolution in foraminiferal assemblages in
order to reduce taphonomic bias in paleoenvironmental reconstructions. To achieve this, we carried out experimental dissolution studies on foraminiferal assemblages from Pliocene–Quaternary sapropel-bearing sequences of Eratosthenes Seamount, Eastern Mediterranean (ODP Leg 160, Site 966). Our experiments refine two general observations on modern foraminifera, which are in fact only applicable for size fraction of 125–630 μm: 1) Dissolution increases the relative abundance of fragmentation in planktic assemblages (%Fp) more rapidly than in benthic assemblages (%Fb); 2) with moderate to severe dissolution, planktic foraminiferal numbers decrease faster than benthic foraminiferal numbers. Besides, our experiments show that dissolution susceptibility differs between architectural types. As dissolution progresses, the proportions of uniloculars, uniserials, biserials, then tri-multiserials and miliolines gradually decrease and benthic foraminiferal assemblages become relatively enriched in planispirals and trochospirals. These findings allow re-evaluation of commonly used foraminiferal dissolution indices. Accordingly, %F can be used to assess dissolution in weakly to moderately dissolved planktic assemblages (size fraction >125 μm) in non-lithified sediments; P/B ratios are a better indicator of dissolution in moderately to severely dissolved assemblages, but only in the size fraction >125 μmand not in the smaller size fraction. Anomalously lowforaminiferal numbers (compared to background numbers in a sequence) can also hint at dissolution. A dominance of planispiral, trochospiral and large taxa in benthic assemblages may be used as an additional dissolution indicator. Based on these results, we propose a new procedure to evaluate dissolution in foraminiferal assemblages. Application of this procedure to some published Pliocene–Quaternary records reveals a good agreement between our experiments and expected consequences of dissolution for these records. In order to achieve more robust interpretations based on quantitative foraminiferal data, the dissolution procedure proposed should be tested elsewhere and could develop into a standard micropaleontologic procedure. This is useful in not only open ocean records, but also continental margin studies dealing with major biotic events or studies employing P/B ratios for sea-level reconstructions are expected to benefit from this approach.