Acute grazer toxicity of freshwater diatom biofilms was determined using Thamnocephalus platyurus Packard, an anostracan grazer, as the bioassay organism. The diatoms exhibited toxicity when the cells of the biofilm were freeze–thawed before the assay. The toxicity could be extracted from the biofilms with methanol and acetone, and only minimal toxicity was left in the insoluble residue. Bioassay-guided HPLC separation of the methanolic extract was performed to trace the most toxic components. Analysis by UV spectrometry, gas chromatography, and mass spectrometry showed that 5,8,11,14,17-eicosapentaenoic acid was responsible for most grazer toxicity. The 24-h LC50 of this polyunsaturated fatty acid was 34 μM in the Thamnocephalus platyurus bioassay. The concentrations of other free fatty acids were not high enough to contribute significantly to the toxicity. Procedures that affected the integrity of the cells (e.g. solvent extraction, freezing and thawing, osmotic stress by addition of 20% NaCl, or grinding the cells in a mortar) were taken as model reactions for grazing and had the common effect of resulting in a dramatic increase of free polyunsaturated and saturated fatty acids. Under these conditions, about 30% of the total fatty acids of the diatoms was transformed from the bound into the free form. The time necessary for liberation was very short. With the exception of 5,8,11,14,17-eicosapentaenoic acid, which continued to be liberated, the hydrolysis of the other fatty acids was terminated less than 1 min after initiating the reaction. The classical extraction procedures using methanol and other solvents led to the appearance of a high percentage of free fatty acids in live cells. Treatment of biofilms with these solvents did not stop the hydrolysis of lipids initiated by the disintegration of the cells. However, boiling acetone completely suppressed the hydrolytic reactions, and free polyunsaturated fatty acids were not detected in live biofilm organisms, although nontoxic saturated fatty acids were present in moderate concentrations. These results were interpreted as an indication that the frequently reported existence of free polyunsaturated fatty acids in live biomass is an analytical artifact.