Opaline phytoliths are important microfossils used for paleoecological and archaeological reconstructions that are primarily based on relative ratios of specific morphotypes. Recent studies have shown that phytolith assemblages are prone to post-depositional alteration involving partial dissolution, however, the manner in which partial dissolution affects morphotype composition is poorly understood. Here we show that morphotype assemblages from four different plant species subjected to controlled partial dissolution are significantly different from the original assemblages, indicating that the stability of various morphotypes differs, mainly depending on their surface area to bulk ratios. This underlying mechanism produces distorted morphotype compositions in partially dissolved phytolith assemblages, bearing vast implications for morphotype-based paleoecological and archaeological interpretation. Together with analyses of phytolith assemblages from a variety of archaeological sites, our results establish criteria by which well-preserved phytolith assemblages can be selected for accurate paleoecological and archaeological reconstructions.