Marine planktonic diatoms export carbon to the deep ocean, playing a key role in the global carbon cycle. Although commonly thought to have diversified over the Cenozoic as global oceans cooled, only two conflicting quantitative reconstructions exist, both from the Neptune deep-sea microfossil occurrences database. Total diversity shows Cenozoic increase but is sample size biased; conventional subsampling shows little net change. We calculate diversity from a separately compiled new diatom species range catalog, and recalculate Neptune subsampled-in-bin diversity using new methods to correct for increasing Cenozoic geographic endemism and decreasing Cenozoic evenness. We find coherent, substantial Cenozoic diversification in both datasets. Many living cold water species, including species important for export productivity, originate only in the latest Miocene or younger. We make a first quantitative comparison of diatom diversity to the global Cenozoic benthic ∂18O (climate) and carbon cycle records (∂13C, and 20-0 Ma pCO2). Warmer climates are strongly correlated with lower diatom diversity (raw: rho = .92, p<.001; detrended, r = .6, p = .01). Diatoms were 20% less diverse in the early late Miocene, when temperatures and pCO2 were only moderately higher than today. Diversity is strongly correlated to both ∂13C and pCO2 over the last 15 my (for both: r>.9, detrended r>.6, all p<.001), but only weakly over the earlier Cenozoic, suggesting increasingly strong linkage of diatom and climate evolution in the Neogene. Our results suggest that many living marine planktonic diatom species may be at risk of extinction in future warm oceans, with an unknown but potentially substantial negative impact on the ocean biologic pump and oceanic carbon sequestration. We cannot however extrapolate our my-scale correlations with generic climate proxies to anthropogenic time-scales of warming without additional species-specific information on proximate ecologic controls.