In order to better understand the maintenance of a fairly narrow diet breadth in monarch butterfly larvae, Danaus plexippus L. (Lepidoptera: Nymphalidae: Danainae), we measured feeding preference and survival on host and non-host plant species, and sensitivity to host and non-host plant chemicals. For the plant species tested, a hierarchy of feeding preferences was observed; only plants from the Asclepiadaceae were more or equally preferred to Asclepias curassavica, the common control. The feeding preferences among plant species within the Asclepiadaceae are similar to published mean cardenolide concentrations. However, since cardenolide data were not collected from individual plants tested, definitive conclusions regarding cardenolide concentrations and plant acceptability cannot be made. Although several non-Asclepiadaceae were eaten in small quantities, all were less preferred to A. curassavica. Additionally, these non-Asclepiadaceae do not support continued feeding, development, and survival of first and fifth-instar larvae. Preference for a host versus a non-host (A. curassavica versus Vinca rosea) increased for A. curassavica reared larvae as compared to diet-reared larvae suggesting plasticity in larval food preferences. Furthermore, host species were significantly preferred over non-host plant species in bioassays using a host plant or sucrose as a common control. Larval responses to pure chemicals were examined in order to determine if host and non-host chemicals stimulate or deter feeding in monarch larvae. We found that larvae were stimulated to feed by some ubiquitous plant chemicals, such as sucrose, inositol, and rutin. In contrast, several non-host plant chemicals deterred feeding: caffeine, apocynin, gossypol, tomatine, atropine, quercitrin, and sinigrin. Additionally the cardenolides digitoxin and ouabain, which are not in milkweed plants, were neutral in their influence on feeding. Another non-milkweed cardenolide, cymarin, significantly deterred feeding. Extracts of A. curassavica leaves were tested in bioassays to determine which components of the leaf stimulate feeding. Both an ethanol extract of whole leaves and a hexane leaf-surface extract are phagostimulatory, suggesting the involvement of both polar and non-polar plant compounds. These data suggest that the host range of D. plexippus larvae is maintained by both feeding stimulatory and deterrent chemicals in host and non-host plants.