Pandemic influenza presents special problems for vaccine development. There must be a balance between rapid availability of vaccine and the safeguards to ensure safety, quality and efficacy of vaccine. Vaccine was developed for the pandemics of 1957, 1968, 1977 and for the pandemic alert of 1976. This experience is compared with that gained in developing vaccines for a possible H5N1 pandemic in 1997-1998. Our ability to mass produce influenza vaccines against a pandemic threat was well illustrated by the production of over 150 million doses of 'swine flu' vaccine in the USA within a 3 month period in 1976. However, there is cause for concern that the lead time to begin vaccine production is likely to be about 7-8 months. Attempts to reduce this time should receive urgent attention. Immunogenicity of vaccines in pandemic situations is compared over the period 1968-1998. A consistent feature of the vaccine trials is the demonstration that one conventional 15 microg haemagglutinin dose of vaccine is not sufficiently immunogenic in naive individuals. Much larger doses or two lower doses are needed to induce satisfactory immunity. There is some evidence that whole-virus vaccines are more immunogenic than split or subunit vaccines, but this needs substantiating by further studies. H5 vaccines appeared to be particularly poor immunogens and there is evidence that an adjuvant may be needed. Prospects for improving the development of pandemic vaccines are discussed.