Hemiparasitic plants have green leaves, but extract water and solutes from neighbouring plants. It is still poorly understood how different host plants in communities contribute to parasite performance, as species that are good hosts in single-host experiments may not necessarily be preferred hosts in mixtures. We grew the root hemiparasite Rhinanthus alectorolophus (Orobanchaceae) together with each of 13 host species (experiment 1) and with 15 different four-species mixtures of these hosts (exp. 2) that differed in the number of legumes and of host functional groups. Parasites profited from mixtures including more legumes and from mixtures including different host functional groups. Some host species and mixtures were very tolerant of parasitism and supported large parasites without being strongly suppressed in their own growth, but the suppression of a species in the single-host experiment did not explain the suppression of a species in a host mixture. We thus calculated for each host species an index of the difference in suppression between the two experiments which may be related to host use in a mixture. Host quality (mean parasite biomass with a host species) in the single-host experiment could explain 64% of the variation in parasite biomass with a host mixture when it was weighted by the proportion of the host species in the mixture without the parasite and by the suppression difference index. Our results suggest that plant species which are the best hosts in single-host experiments are not always those used most strongly by a parasite growing with a mixture. Together with the finding that hemiparasites benefit from a mixed diet based on hosts from different functional groups this suggests that parasites prefer certain host species to obtain a mixed diet.