The SeaWiFS potential for identifying the marine cyanobacterium Trichodesmium, at low concentration in oligotrophic case I waters, has been investigated by numerical simulations using a three-component model of sea colour. The results obtained indicate that simple algorithms, which operate on data recorded by SeaWiFS bands 2, 3. and 5, should have the capability to discriminate pixels containingTrichodesmiumat concentrations as low as 0·1–0·3 mgm-3chlorophyll a, as well as to yield an estimate of the concentration. The validity range of the procedure has been explored by a sensitivity analysis, assuming realistic variations in the input data to the optical model.