Successful use of Glyptapanteles militaris (Walsh) to control the polyphagous pest Pseudaletia unipuncta (Haworth) depends on knowledge of the host–parasite relationship. In this study, we determined the suitability of different P. unipuncta instars for oviposition by G. militaris females, and we evaluated the parasitoid performance on each host instar (L1 to L6). The following parameters were measured: percentage of larvae parasitized from which parasitoids emerged; percentage of host mortality; percentage of hosts that pupated after being stung by the parasitoid; egg–larval developmental time; pupal period; total developmental time; mean number of cocoons per host; mean number of parasitoid larvae that emerged from a host larva but failed to spin a cocoon; number of parasitoid larvae that failed to emerge from each host; total number of parasitoid larvae per host; parasitoid sex ratio and emergence rate of adult progeny. All host instars were stung by female parasitoids, but cocoons were never obtained from L6. The proportion of parasitized larvae increased from L1 to L4 then declined steeply, reaching no parasitism for L6, with a significant difference between L1 and L3, L4. Conversely, the proportion of host larvae that pupated decreased from L1 to L3, increased for L4 and L5, and slightly declined for L6. Significant differences were observed between L5, L6 and the other 4 instars. Similar results were observed for the proportion of larva killed before emergence of the parasites, with a significant difference between L6 and L2, L3. Host instar at the time of parasitization significantly affected egg–larval and total developmental time, the total number of parasitoid larvae per host, and the emergence rate of adult progeny. These results indicate that the 2nd, 3rd, and 4th instars of P. unipuncta were more suitable for G. militaris parasitism, because these instars provide a high percentage of hosts that will be successfully parasitized and they also produce large numbers of parasitoids per host.