We present a quantitative method, based on the relative abundances of benthic foraminifera in deep-sea sediments, for estimating surface ocean biological productivity over the timescale of centuries to millennia. We calibrate the method using a global data set composed of 207 samples from the Atlantic, Pacific, and Indian Oceans from a water depth range between 2300 and 3600 m. The sample set was developed so that other, potentially significant, environmental variables would be uncorrelated to overlying surface ocean productivity. A regression of assemblages against productivity yielded an r2=0.89 demonstrating a strong productivity signal in the faunal data. In addition, we examined assemblage response to annual variability in biological productivity (seasonality). Our data set included a range of seasonalities which we quantified into a seasonality index using the pigment color bands from the coastal zone color scanner (CZCS). The response of benthic foraminiferal assemblage composition to our seasonality index was tested with regression analysis. We obtained a statistically highly significant r2=0.75. Further, discriminant function analysis revealed a clear separation among sample groups based on surface ocean productivity and our seasonality index. Finally, we tested the response of benthic foraminiferal assemblages to three different modes of seasonality. We observed a distinct separation of our samples into groups representing low seasonal variability, strong seasonality with a single main productivity event in the year, and strong seasonality with multiple productivity events in the year. Reconstructing surface ocean biological productivity with benthic foraminifera will aid in modeling marine biogeochemical cycles. Also, estimating mode and range of annual seasonality will provide insight to changing oceanic processes, allowing the examination of the mechanisms causing changes in the marine biotic system over time. ¿ 1999 American Geophysical Union