One of the foraging decisions facing parasitoids is whether to accept (superparasitize) or reject hosts that have already been parasitized. An important distinction is whether the host has been parasitized by the female parasitoid herself or by a conspecific. In solitary parasitoids, the pay-off from an egg laid in the latter host type (conspecific superparasitism) is the probability that the second egg wins the competition for the host and results in an offspring. The pay-off from an egg laid in the former type (self-superparasitism) increases with an increasing probability that another female will superparasitize the host in the near future. When this probability equals one, self-superparasitism and conspecific superparasitism have the same payoff. However, conspecific superparasitism will generally have a higher pay-off than self-superparasitism. It will therefore be beneficial for a female parasitoid to be able to distinguish between a host she parasitized and one parasitized by a conspecific. The degree of benefit depends on the probability of conspecific superparasitism in the near future. Using an optimal diet model, I show that when a parasitoid encounters a patch containing a mixture of unparasitized and already-parasitized hosts, a female that can distinguish between the two types of parasitized hosts gains more offspring than a female without this ability. However, when parasitoids search a patch together with conspecifics, it is adaptive to self-superparasitize, and the pay-off from this ability may be negligible. It is therefore predicted that when a female parasitoid searches a partially depleted patch alone, it will reject the hosts parasitized by itself more frequently than hosts parasitized by conspecifics. In contrast, female parasitoids searching together are predicted to accept hosts that they parasitized themselves much more often. The results show that the solitary parasitoid Leptopilina heterotoma (Hymenoptera: Eucoilidae) is able to distinguish between hosts that it parasitized and hosts parasitized by conspecifics. The predictions of the model are met in a second experiment that shows that L. heterotoma self-superparasitizes when the probability of conspecific superparasitism is high.