A survey of nurses at the Prince of Wales Children's Hospital was conducted to determine the prevalence of immunity to varicella-zoster virus (VZ) as defined by enzyme immunoassay (EIA), and to establish the value of history as a predictor of immunity. Of the 209 nurses surveyed, 51% could recall suffering VZ infection, and with a single exception, all of this group were immune. However, despite a 95% prevalence of immunity among all nurses, 46% of those found to be immune by EIA could not recollect having VZ infection. In the event of a hospital VZ outbreak, the latter group, without serological testing, would thus need to be regarded as susceptible, and this would create a major logistical problem in staffing the affected areas. We suggest, to minimize this cause of disruption to services, that all paediatric staff with patient contact should be asked at the time of recruitment if they recall suffering VZ infection. Those who give a positive response may be considered immune, but all other staff should have their immune status assessed by EIA at the earliest opportunity.