AbstractPaleoceanographic studies often rely on abundance changes in microfossil species, with little consideration for characteristics such as organism size, which may also be related to environmental changes. Using a tropical Indian Ocean (TIO) core‐top data set, we test the Optimum size‐hypothesis (OSH), investigating whether relative abundance or environmental variables are better descriptors of planktonic foraminifera species' optimum conditions. We also investigate the environmental drivers of whole‐assemblage planktonic foraminiferal test size variation in the TIO. We use an automated imaging and sorting system (MiSo) to identify planktonic foraminiferal species, analyze their morphology, and quantify fragmentation rate using machine learning techniques. Machine model accuracy is confirmed by comparison with human classifiers (97% accuracy). Data for 33 environmental parameters were extracted from modern databases and, through exploratory factor analysis and regression models, we explore relationships between planktonic foraminiferal size and oceanographic parameters in the TIO. Results show that the size frequency distribution of most planktonic foraminifera species is unimodal, with some larger species showing multimodal distributions. Assemblage size95/5 (95th percentile size) increases with increasing species diversity, and this is attributed to vertical niche separation induced by thermal stratification. Our test for the OSH reveals that relative abundance is not a good predictor of species' optima and within‐species size95/5 response to environmental parameters is species‐specific, with parameters related to carbonate ion concentration, temperature, and salinity being primary drivers. At the species and assemblage levels, our analyses indicate that carbonate ion concentration and temperature play important roles in determining size trends in TIO planktonic foraminifera.