Net primary production (NPP) by microscopic phytoplankton underpins nearly all marine life, yet global NPP estimates differ substantially. Among satellite-based estimates, variation has been attributed to differing assumptions about the relationships between photosynthesis and sea surface temperature (SST). Maximum chlorophyll (Chl)–specific rates of carbon fixation in the water column (PBopt) increase monotonically with SST in some satellite algorithms, whereas others peak at intermediate values. Understanding and constraining such relationships is challenged by the many direct and indirect relationships between temperature and phytoplankton. In this article, the emergent PBopt-SST relationship was diagnosed in a global biogeochemical simulation and compared with widely used satellite NPP algorithms. The simulated PBopt-SST relationship for the aggregated phytoplankton community was highly significant (r2= 0.83) and increased monotonically with temperature. The PBopt-SST relationships for small and large phytoplankton, however, were distinct and weaker than the aggregate relationship (r2 = 0.52 and 0.36, respectively). For small phytoplankton, the inhibitory effect of nutrient limitation in warmer, more stratified waters was moderated by efficient nutrient scavenging. This, combined with photoacclimation and the stimulatory effect of warming on maximum growth produced steep PBopt increases with SST. For large phytoplankton, the need for higher Chl:C in productive regions (because of package effects) and the onset of severe nutrient limitation in warm, stratified regions decreased PBopt relative to small phytoplankton, though PBopt still increased modestly with SST. The PBopt-SST relationships for both small and large phytoplankton overestimated PBopt in nutrient-poor regions and underestimated it in nutrient-rich regions. This bias was greatly reduced in the aggregate relationship because the increased prominence of low-PBopt large phytoplankton in nutrient-rich environments tempered PBopt increases. Results support a strong, monotonically increasing PBopt-SST relationship but emphasize the role of the size structure of the phytoplankton community in shaping the emergent relationship. The implications of these results are discussed in relation to efforts to improve satellite NPP algorithms and partition NPP by phytoplankton size.