The carbohydrates expressed on the surface of meningococcal strains of groups B and C mimic those commonly found on human cells and thus are not functionally antigenic in infancy. In order to develop an effective vaccine, it will be necessary to find ways of circumventing this molecular mimicry. Three possible ways of achieving this are discussed. (i) The surface polysaccharides can theoretically present conformationally different epitopes, some of which might be recognized as antigenic by the host. Experimental evidence is presented that such differences do indeed exist; what is needed is to determine which of these conformations are unique to the organism and hence potentially antigenic. (ii) Precursors of the surface lipooligosaccharides may be unable to mimic human antigens, and so may be potential candidates for vaccine development. (iii) Natural immunity to some strains of meningococci develops in young children who are colonized with strains of Neisseria lactamica, and it is possible that its development could be enhanced by widespread intentional colonization by N. lactamica strains that are particularly efficient inducers of broad immunity.