Defective interfering (DI) virus RNAs result from major deletions in full-length viral RNAs that occur spontaneously during de novo RNA synthesis. These RNAs are packaged into virions that are by definition non-infectious, and are delivered to cells normally targeted by the virion. DI RNAs can only replicate with the aid of a coinfecting infectious helper virus, but the small size of DI RNA allows more copies of it to be made than of its full-length counterpart, so the cell produces defective virions in place of infectious progeny. In line with this scenario, the expected lethal disease in an influenza A virus-mouse model is made subclinical by administration of DI virus, but animals develop solid immunity to the infecting virus. Hence DI virus has been called an ´interfering vaccine´. Because interfering vaccine acts intracellularly and at a molecular level, it should be effective against all influenza A viruses regardless of subtype. Here we have used the ferret, widely acknowledged as the best model for human influenza. We show that an interfering vaccine with defective RNAs from an H3N8 virus almost completely abolished clinical disease caused by A/Sydney/5/97 (H3N2), with abrogation of fever and significant reductions in clinical signs of illness. Animals recovered fully and were solidly immune to reinfection, in line with the view that treatment converts the otherwise virulent disease into a subclinical and immunizing infection.