Hyperparasites (species which parasitise other parasites) are common in natural populations and affect many parasitic taxa, including: eukaryotic parasites; bacterial and fungal pathogens; and insect parasitoids. Hyperparasitism is therefore likely to shape the ecology and evolution of many host-parasite systems, and represents a promising method for biocontrol (e.g. treating antimicrobial resistant infections). However, the eco-evolutionary consequences of hyperparasitism have received little attention. We use a host-parasite-hyperparasite model to explore how introducing a hyperparasite drives the evolution of parasite virulence, and how this affects host population dynamics. We show when the introduction of a hyperparasite selects for higher or lower parasite virulence, and how this changes the disease burden for the host population. Crucially, we show that variation in the virulence and infectivity of hyperparasites, along with the probability of co-transmission, can lead to a previously unseen hysteresis effect, whereby small shifts in hyperparasite characteristics can lead to sudden shifts in parasite virulence. We show that hyperparasites can induce diversification in parasite virulence, leading to the coexistence of high and low virulence strains. Our results show hyperparasites can have dramatic effects on the evolution of parasite virulence, and that the use of hyperparasites in biocontrol should therefore be approached with caution.