Ivermectin has potent systemic activity against numerous species of nematodes and arthropods, but there are some important species in these two groups, such as the cat flea, Ctenocephalides felis (Bouché), that appear to be refractory to it. In an effort to determine if the lack of systemic activity against C. felis is specific to ivermectin, or if it is a class-wide phenomenon, 20 avermectin derivatives were tested in an artificial membrane flea feeding system at concentrations of 20, 10, and 1 μg/ml. Results showed that ivermectin had LC₉₀ and LC₅₀ values against fleas of 19.1 and 9.9 μg/ml, respectively. Only four of the other 19 compounds evaluated possessed both LC₉₀ and LC₅₀ values more potent than ivermectin and even then the advantage was modest. Among those four compounds was a two-fold increase in potency relative to ivermectin when the LC₉₀ values were considered (range, 9.2–10.3 μg/ml) and a two- to eight-fold increase when the LC₅₀ values were examined (range, 1.23–5.26 μg/ml). Neither the possession nor the number of oleandrosyl sugars on the macrocyclic backbone were relevant for additional flea activity because among these four compounds were two disaccharides, a monosaccharide and an aglycone. Also, bond disposition between C-22 and 23 did not contribute to increase in activity because these molecules comprise members with either single or double bonds. One of these avermectin analogs was scaled-up and tested subcutaneously in a dog at >100 times the commercial ivermectin dosage and zero efficacy was observed against the flea. We conclude that even the best in vitro avermectin does not have the in vivo potential to become a commercial oral or subcutaneous flea treatment for companion animals.