The toxicity of a series of 120 aliphatic alcohols was evaluated using the Tetrahymena pyriformis population growth impairment assay. For tertiary propargylic alcohols; primary, secondary, and tertiary homopropargylic alcohols; allylic alcohols; and saturated alcohols, a statistically robust structure-activity model was developed for toxicity data [log (IGC(50) (-1))] using the 1-octanol/water partition coefficient (log K(ow)) as the lone descriptor [log (IGC(50))(-1) = 0.74 (log K(ow)) - 1.73; n = 97; r(2) (adj.) = 0.933; r(2) (pred.) = 0.932; s = 0.298; F = 1328; Pr > F = 0.0001]. Analysis of data for the primary propargylic alcohols yielded a separate, high-quality log K(ow)-dependent quantitative structure-activity relationship (QSAR) [log (IGC(50))(-1) = 0.65 (log K(ow)) - 1.22; n = 10; r(2) (adj.) = 0.969; r(2) (pred.) = 0.964; s = 0.222; F = 254; Pr > F = 0.0001]. A comparison of the observed toxicity and that predicted by the first QSAR showed that the primary propargylic alcohols with log K(ow) values < 2.00 exhibited enhanced toxicity and that this increased toxicity was inversely related to hydrophobicity. In sharp contrast, analysis of the data for the secondary propargylic alcohols exhibited little relationship with log K(ow) (r(2) = 0.339). Although the initial QSAR can be used to model the toxicity of any aliphatic alcohol for the T. pyriformis population growth impairment end point, the estimated potency would be underestimated for primary propargylic alcohols with log K(ow) values < 2.00. Moreover, estimates of toxic potency of secondary propargylic alcohols based on this QSAR should be viewed with limited confidence. The findings for beta-unsaturated alcohols in Tetrahymena were sharply different from that reported for fathead minnow acute mortality; this difference in toxicity is a result of a difference in the protocol used rather than in metabolism.