We present a deterministic model of the possible emergence of a vaccine escape variant of hepatitis B virus (HBV). The model identifies the key unknowns determining this process: the protection afforded by the current vaccines against particular HBV variants; the infectiousness of these variants; and the current prevalence of individuals infectious with the variants (each factor relative to wild-type). By making pessimistic assumptions about these unknowns we show that even a highly infectious variant, under a vaccine programme that affords no protection against the variant, would still take decades to emerge. Thus the current low prevalence of variants is not evidence for the cross-reactivity of the current vaccines or for a lack of infectiousness in the variants. As any vaccine failure will be inapparent for decades it may be sensible to recommend vaccine modifications now rather than later.