Up to the present time the polyhedral viruses of insects have been considered to be extremely specific in their infectivity, and cases of cross-infection, even between closely related species, were regarded as rare occurrences. Recently, however, we have recorded results which suggest that there is no such specificity of infection (Smith & Xeros, 1952). In the following paper we describe the results obtained in an extensive series of cross-inoculations among a heterogeneous collection of lepidopterous larvae, selected entirely at random. The results obtained appear at first sight to be compatible with one clear-cut explanation of straightforward cross-infections with differential multiplication of two different viruses in different hosts. Other complications, however, cannot be excluded and one or two possible situations are discussed later in the paper.