Vaccine escape variants of hepatitis B virus (HBV) have been identified world-wide. A mathematical model of HBV transmission is used to investigate the potential pattern of emergence of such variants. Attention is focused on The Gambia as a country with high quality epidemiological data, universal infant immunization and in which escape mutants after childhood infections have been observed. We predict that a variant cannot become dominant for at least 20 years from the start of vaccination, even when using a vaccine which affords no cross protection. The dominant factor responsible for this long time scale is the low rate of infectious contacts between infected and susceptible individuals (we estimate the basic reproduction number of hepatitis B in The Gambia to be 1·7). A variant strain that achieves high prevalence will also take many years to control, and it is questionable whether emergence will be identifiable by sero-surveillance until of high prevalence. The sensitivity of the model predictions to epidemiological and demographic factors is explored.