The mechanism of cross-protection, the deliberate infection of plants with a “mild” virus isolate to protect against “severe” isolates, has long been a topic of debate. In our model system, Citrus tristeza virus (CTV), this appears to be genotype-specific superinfection-exclusion, suggesting a simple recipe for cross-protection. However, this concept failed in field trials, which led us to examine the process of superinfection-exclusion more closely. We found that exclusion relies on the relative fitness of the primary versus the challenge isolates, and the host infected, and that significant differences in superinfection success could occur between isolates that differ by as few as 3 nucleotides. Furthermore, we found that exclusion was not uniform throughout the plant, but was tissue-specific. These data suggest that cross-protection is not a simple like-for-like process but a complex interaction between the primary and challenge isolates and the host. •Superinfection-exclusion is not a simple genotype-specific like-for-like process.•Isolate fitness and host species determine the outcome of superinfection.•Even small changes of a few nucleotides can alter the success or failure of superinfection.