Phytoplankton blooms are uncoupled from grazing and are normally terminated by sedimentation. There are several potential mechanisms by which phytoplankton cells may settle out of the photic zone: sinking of individual cells or chains, coagulation of cells into aggregates with high settling velocities, settling of cells attached to marine snow aggregates formed from discarded larvacean houses or pteropod feeding webs, and packaging of cells into rapidly falling zooplankton fecal pellets. We quantified the relative significance of these different mechanisms during a diatom bloom in a temperate fjord, and evaluated their potential to control phytoplankton population dynamics. Overall specific sedimentation rates of intact phytoplankton cells were low during the 11-day study period, averaging ca. 0.1 d−1, and mass sedimentation and bloom termination did not occur. Most cells settled attached to marine snow aggregates formed from discarded larvacean houses, whereas settling of unaggregated cells was insignificant. Formation rates of phytoplankton aggregates by physical coagulation was very low, and losses by this mechanism were <0.07 d−1; phytoplankton aggregates were neither recorded in the water column (by divers) nor in sediment traps. The low coagulation rates were due to a very low ‘stickiness’ of suspended particles. The dominant diatom, Thalassiosira mendiolana, that accounted for up to 75% of the phytoplankton biomass, was not sticky at all, and did not turn sticky upon nutrient depletion in culture experiments. The low particle stickiness recorded may be related to low formation rates by diatoms of transparent exopolymeric particles (TEP), that occurred in low concentrations throughout the study period. Zooplankton grazing rate did not respond to the development of the bloom and accounted for a loss term to the phytoplankton populations comparable to sinking of intact cells; fecal pellets accounted for 30–50% of settled phytoplankton and phytodetritus. While coagulation may give rise to density-dependent losses to phytoplankton populations and, hence, control blooms, neither of the other mechanisms examined worked in a density dependent manner. In the absence of significant coagulation rates, rapid mass sedimentation of this bloom did not occur.