Preparations of fresh or dried feverfew (Chrysanthemum parthenium) are widely consumed in the U.K. as a remedy for arthritis and migraine, but the pharmacological basis for this has not been established. We have, therefore, compared the properties of extracts of fresh plants with those of dried powdered leaves available commercially from health food shops. The two extracts differed radically in their content of alpha-methylbutyrolactones and in their pharmacological profile when tested in vitro on the rabbit aortic ring and rat anococcygeus preparations. Extracts of fresh leaves caused dose- and time-dependent inhibition of the contractile responses of aortic rings to all receptor-acting agonists so far tested; the effects were irreversible and may represent a toxic modification of post-receptor contractile function in the smooth muscle. The presence of potentially-SH reactive parthenolide and other sesquiterpene alpha-methylenebutyrolactones in these extracts, and the close parallelism of the actions of pure parthenolide, suggest that the inhibitory effects are due to these compounds. In contrast, chloroform extracts of dried powdered leaves were not inhibitory but themselves elicited potent and sustained contractions of aortic smooth muscle that were not antagonised by ketanserin (5-HT2 receptor antagonist). These extracts did not contain parthenolide or butyrolactones according to a chemical-HPLC assay. We conclude that there are marked differences in the pharmacological potency and profiles between preparations from fresh and dried feverfew and that this may relate to their lactone content. As the effects of the lactones are potentially toxic, it will be necessary to compare the clinical profiles and side effects of preparations obtained from the two sources.