A comparative challenge study of six marine isolates representing subtypes 1–6 of salmonid alphavirus (salmon pancreas disease virus, Genus Alphavirus, Family Togaviridae) was conducted in Atlantic salmon in a fresh water cohabitation trial. Histopathological lesions typical of pancreas disease were observed with all subtypes, and virus was re-isolated from serum of cohabitant fish in each case. Using a virus neutralization (VN) test neutralizing salmonid alphavirus (SAV) subtype 1 strain F93-125, VN antibodies were detected in all challenge groups, consistent with serological cross-reactivity between these subtypes. Using real-time RT-PCR, SAV RNA was detected in heart tissue from 2 to 3 weeks post-challenge (wpc) in all cohabitant groups excluding controls. The results obtained suggested differences in the dynamics of infection between strains of SAV and potentially between subtypes. Results for SAV subtypes 1 and 3 suggested essentially synchronous infection of cohabitant fish. These two study groups also had the highest virus load in heart tissue as measured by quantitative RT-PCR and also had the most extensive histopathological changes. In contrast, results for SAV subtypes 2 and 6 strains were consistent with asynchronous infection in the cohabitant fish and were characterized by slow spread, low virus loads and mild histopathological changes. The SAV subtype 4 and 5 strains occupied an intermediate position in this regard. Despite the use of concentration procedures, it was not possible to detect SAV RNA in water samples from selected study tanks. However, testing of faeces from the SAV subtypes 1, 3 and 6 challenge groups found positive signals in each beginning at 1–3 wpc and remaining detectable for a further 2–3 weeks. Parallel testing of mucus samples found these became positive at 2–3 wpc and remained positive for a further 1–3 weeks. These results demonstrate for the first time that shedding and transmission of virus may occur by both these routes and suggest that dispersal in these matrices should be included in any disease transmission models.