Knowing which fraction of a phytoplankton population is viable would often be helpful in answering ecological or physiological questions. However, viability stains (1) often do not function properly, especially with diatoms, (2) are rarely used, and (3) frequently appear to give ambiguous results. Here, we investigate the performance of the FDA viability assay in detail and test its applicability for investigating 2 ecologically important questions. The functioning of the assay was tested for several diatom species, as well as for Emiliania huxleyi (coccolithophorid) and Phaeocycstis globosa (Prymnesiophyceae). Additionally, changes in the viability of Thalassiosira weissflogii (diatom)--(1) due to aggregation and (2) during a prolonged stationary phase induced by nutrient limitation--were explored. Whereas the method generated solid results with some species, including T. weissflogii, it was less trustworthy or even totally ineffective with others. Ambiguous definitions of viability and cell death, as well as temporary changes in the strength of the fluorescence signal, complicate the interpretation of viability measurements. However, exiting ecological results can be attained with this method. Experiments with T. weissflogii suggested that the vast majority of cells, rather than individual survivors remained viable for over a month in an illuminated, nutrient-devoid environment. Aggregation significantly prolonged the viability of T. weissflogii cells compared to non-aggregated cells kept under otherwise identical conditions.