Three trials involving 184 male Sprague Dawley rats were conducted to identify the biologically active fraction of endophyte-infected tall fescue seed. In trial 1, seed infected (greater than 95%) with the endophytic fungus Acremonium coenophialum and shown to be toxic to cattle, was serially extracted with hexane, ethylacetate, and methanol, respectively. A second sample of this seed was extracted with chloroform only. Controls included a solvent treatment (spraying the various solvents on endophyte-free seed and carrying the mixture to dryness), the seed post-extraction, endophyte-infected seed and endo-phyte-free seed. Laboratory chow was mixed with all diets to provide 50% of the mix. Trial 2 examined the same treatments as trial 1 except that the chloroform treatment was omitted. Dosages 3 times (equivalent to 22.5 g seed) those used in trial 1 were applied to endophyte-free seed (7.5 g) in an attempt to accentuate any differences. Trial 3 examined the methanol fraction extracted serially, a methanol fraction extracted batchwise, and a water fraction extracted batchwise. Controls included seed residues post-extraction, solvent residue from methanol sprayed on endophyte-free seed, and endophyte-infected seed. Lowered feed intake and weight gains, and depressed serum prolactin concentrations were used as indicators of toxicity. Feed intake was lower in all three trials for rats fed the methanol fraction. In all three trials mean serum prolactin concentrations in rats fed the methanol extracts were not different from those fed endophyte-infected fescue seed. Rats fed the water extract in trial 3 had lower feed intakes than rats fed endophyte-free seed. However, rats fed the seed, post-water extraction, had serum prolactin concentrations similar to that of the negative control. Results suggest that the toxic factor(s) may be concentrated in the methanol extract.