Rabbitfish Siganus fuscescens preferences for Lyngbya majuscula collected from three bloom locations in Moreton Bay, Queensland, Australia, were tested along with a range of local plant species in the laboratory. Consumption of L. majuscula by fish did not differ between wild and captive-bred fish (P?=?0籵2) but did differ between bloom location (P?=?0簳9). No relationship was found between consumption rates and lyngbyatoxin-a concentration (r2?=?0簳5, P? =?0縱4). No correlation existed between C?:?N and proportion of food consumed when all food types were analysed statistically, whereas a clear correlation was observed when L. majuscula was removed from the calculations. In simulated bloom conditions, fish avoided ingestion of L. majuscula by feeding through gaps in the L. majuscula coverage. Both wild and captive-bred S. fuscescens showed a distinct feeding pattern in 10 day no-choice feeding assays, with less L. majuscula being consumed than the preferred red alga Acanthophora spicifera. Lyngbya majuscula however, was consumed in equal quantities to A. spicifera by wild S. fuscescens when lyngbyatoxin-a was not detectable. Wild fish probably do not preferentially feed on L. majuscula when secondary metabolites are present and are not severely impacted by large L. majuscula blooms in Moreton Bay. Furthermore, poor feeding performance in both captive-bred and wild S. fuscescens suggests that they would exert little pressure as a top-down control agent of toxic L. majuscula blooms within Moreton Bay.