AbstractIt is well documented that partial dissolution of planktic foraminiferal tests results in a reduction of Mg/Ca ratios, and hence of inferred calcification temperatures; however, traditional analysis techniques have made it difficult to identify the exact mechanism through which Mg is lost. Three hypotheses have been proposed as models for Mg loss for a given extent of dissolution: (1) a percent loss of Mg in individuals, (2) a molar loss of Mg in individuals, and (3) a loss of the highest‐Mg (warmest) individuals from a population. It is vital to better constrain these models as they have very different implications for Mg/Ca paleotemperature dissolution corrections. Here we use a novel individual foraminifera Mg/Ca method to examine the effects of dissolution on the Mg/Ca paleothermometer in three species of planktic foraminifera, Globigerinoides ruber, Neogloboquadrina dutertrei, and Pulleniatina obliquiloculata, from a depth transect of core tops on the Ontong Java Plateau in the western equatorial Pacific. With the exception of the most heavily dissolved population of P. obliquiloculata, our data best support a percent Mg loss model as indicated by the preservation of inferred temperature distribution shapes among the sampled populations and the close fit of the simulated percent Mg loss model to the observed data. Coupled with estimates for foraminiferal dissolution, identification of the percent Mg loss model will allow for more accurate dissolution corrections in Mg/Ca paleothermometry work.