What sets CRISPR-Cas systems apart from other prokaryotic anti-phage defense systems, and gives it such versatility as a genome-editing tool, is the adaptive nature of the immunity conferred. The process of acquiring CRISPR-based immunity against virulent phages requires exposure of the host cell to the phage genome – an inherently risky proposition – in order to sample it for protective ‘spacers’. We hypothesized that the majority of spacer acquisition happens upon infection by a defective phage, either incapable of replication or significantly delayed in its ability to damage the host cell beyond recovery. By challenging Strepocococcus thermophilus cells with phages rendered defective either by a restriction-modification system or by UV irradiation, we demonstrated that cells readily acquire spacers from defective phages. Furthermore, the rate of CRISPR-mediated survival is directly proportional to the quantity of replication-deficient phages to which the cells are exposed. This is reminiscent of immunization in humans through vaccination with inactivated or attenuated viruses. We then sought to refine this immunization process to not only increase rates of acquisition, but bias (and select for) acquisition events in order to ensure specific DNA sequences would be targeted. Using plasmids, analogous to wholly defective phages, as (transient) ‘programming’ vectors, we were able to specify the sequence acquired by the natural adaptation process. This is akin to epitope-specific vaccination. By targeting conserved phage sequences, it allows the generation of strains immune to phages they have not even encountered yet. It can also be used to generate strains refractory to undesired plasmids or antibiotic resistance genes.