In both field samples and cultures the total amount of sugar and of carbon in colonies of Phaeocystis globosa was correlated with colony surface area, suggesting a hollow structure. A conceptual model based on biochemical data and on the assumption that the mucus occurs as a layer of a fixed thickness, irrespective of colony size, predicts that the thickness of the layer is 7 μm. A confocal laser scanning microscope image of fluorescently labelled mucus confirmed this view of the colony structure. The measured contents of carbon and sugar per cell (including mucus) were constant for all colony sizes. Cells in laboratory cultures contained 122 pg C, which is twice the value for cells obtained at a field station in the North Sea (57 pg). In contrast, sugar per cell was higher in the field than in exponentially growing cultures. Therefore the percentage of sugar carbon relative to total carbon of colonies in the field was higher (19–35%) than that in cultures (10%). Highest values were found at low ambient nutrient concentrations, probably due to the presence of storage glucans in the cells that are produced under nutrient limitation. The functional role of Phaeocystis colonies in the pelagic system has to be re-evaluated because (i) biomass estimates have previously often been based on the assumption of a colony filled with mucus while in reality it is a hollow structure with an aqueous lumen, and (ii) in contrast to earlier reports it is shown that only a small amount of carbohydrate is needed to build the colony structure.