Frogs, toads, and salamanders are well known for harboring an array of distasteful (and poisonous) secondary metabolites, presumably as antipredator defenses; yet few experiments have rigorously demonstrated the efficacy of amphibian chemical defenses against ecologically relevant consumers. For example, despite an absence of rigorous statistical evidence showing their distastefulness to predators, eastern newts (Notophthalmus viridescens (Rafinesque))--a common salamander in lentic North American habitats--are assumed to tolerate diverse predator assemblages because newts secrete tetrodotoxin (TTX), a neurotoxin. Here we combine laboratory and field-based ecology with bioassay-guided separation of chemical extracts to show that eastern newts--although chemically protected against ecologically important consumers in lentic systems--nonetheless suffer substantial predation when tethered in the field. When offered newts with alternative prey (paedomorphic Ambystoma talpoideum), red swamp crayfish (Procambarus clarkii) and largemouth bass (Micropterus salmoides) were 9-10x as likely to feed on A. talpoideum as newts. Additionally, juvenile bluegill (Lepomis machrochirus) were 70% less likely to consume newt eggs compared to control food pellets. We also show that different newt tissues were differentially palatable to predatory fish. All bluegill tested consumed a palatable control food, but only 20% consumed dorsal skin, only 35% ate ventral skin, but 75% fed on newt viscera, suggesting that deterrent metabolites are concentrated in the skin. Bioassay-guided fractionation revealed that crude and water-soluble newt chemical extracts inhibited bluegill feeding, definitively establishing the chemical nature of newt antipredator defenses, although we were unsuccessful at isolating the chemical compounds responsible for unpalatability. Yet, deterrent activity in the polar but not the lipophilic chemical fraction and bioassay results demonstrating that naıve predators rapidly learn to avoid natural concentrations of TTX support the possible role of TTX in suppressing predation on newts. However, when tethered in the field, newt mortality was 55% higher in ponds with predatory fishes than in ponds lacking fishes (62% vs. 40% respectively), indicating the possible existence of other predators that are resistant to (or tolerant of) newt chemical defenses. Together, these results stress the importance of rigorous, ecologically relevant, and hypothesis-driven experimentation to better understand the complexity of chemically- mediated predator-prey interactions, even for well-studied species like N. viridescens.