1. Many animals that consume freshwater macrophytes are omnivorous (i.e., they include both plant and animal matter in their diet). For invertebrate omnivorous consumers, selection of macrophyte species depends partly on the presence of secondary metabolites in plants, plant carbon/nutrient balances and/or physical structure of plants. However, little is known about the mechanisms influencing consumption of macrophytes in aquatic vertebrates. 2. For two fish species, the omnivorous rudd (Scardinius erythrophthalmus) and herbivorous grass carp (Ctenopharyngodon idella), feeding preferences were determined in three choice experiments. We tested (i) whether the presence of secondary metabolites and macrophyte stoichiometry affects macrophyte species selection by fish, (ii) the importance of macrophyte stoichiometry by manipulating the macrophytes experimentally and (iii) the rate of herbivory when the most palatable macrophyte is offered simultaneously with a common animal prey. 3. In a choice experiment with five species of submerged macrophytes (Callitriche sp., Chara globularis, Elodea nuttallii, Myriophyllum spicatum and Potamogeton pectinatus), Myriophyllum was clearly consumed least by both fishes, which strongly correlated with the highest phenolic concentration of this macrophyte. Additionally, a significant negative relationship was found between consumption and C : N ratio of the five macrophytes. The two most consumed macrophytes also had the lowest dry matter concentration (DMC). 4. In a second choice experiment, the C : N ratio of the least (Myriophyllum) and most (Potamogeton) palatable plants was manipulated by growing the macrophytes under fertilised and unfertilised conditions and subsequently feeding them to rudd. The avoidance of consumption of the chemically defended Myriophyllum by rudd was partly alleviated by the lowered C : N ratio. 5. The third choice experiment showed that both fishes preferred animal prey (the amphipod Gammarus pulex) over the most palatable macrophyte (Potamogeton) when offered simultaneously. The C : N ratio of the amphipods was about half that of the lowest C : N ratio measured in the macrophytes. Consumption by the fishes could not clearly be related to C : P or N : P ratios of prey items in any of the experiments. 6. We conclude that omnivorous fish avoid macrophytes that are chemically defended. However, when these defences are only minor, stoichiometry (C : N ratio) in combination with DMC may be a determining factor for consumption by vertebrate facultative herbivores.