Phytoplankton account for roughly half of global primary production; it is vital that we understand the processes that control their abundance. A key process is cell division. We have, however, been unable to estimate division rate in natural populations at the appropriate timescale (hours to days) for extended periods of time (months to years). For phytoplankton, the diel change in cell size distribution is related to division rate, which offers an avenue to obtain estimates from in situ observations. We show that a matrix population model, fit to hourly cell size distributions, accurately estimates division rates of both cultured and natural populations of Synechococcus. Application of the model to Synechococcus at the Martha's Vineyard Coastal Observatory provides an unprecedented view that reveals a distinct seasonality in division rates. This information allows us to separate the effects of growth and loss quantitatively over an entire seasonal cycle. We find that division and loss processes are tightly coupled throughout the year. The large seasonal changes in cell abundance are the result of periods of time (weeks to months) when there are small systematic differences that favor either net growth or loss. We also find that temperature plays a critical role in limiting division rate during the annual spring bloom. This approach opens a path to quantify the role of Synechococcus in ecological and biogeochemical processes in natural systems.