The prospects are reviewed of replacement of malaria vector populations by harmless mosquito populations by means of: (i) ecologically competitive non-vector species; (ii) natural selection due to the harmfulness of being infected; (iii) selection for insecticide resistance genes; (iv) meiotic drive; (v) negative heterosis; and (vi) hybrid dysgenesis. Serious difficulties exist with all of these approaches. At present 'dilution', i.e. release of insects carrying the desired genes without any system for forcing population replacement is the only available method. It avoids the disadvantage that, in constructing elaborate genetic 'packages', factors for low fitness may be irreversibly incorporated into them. It is debatable whether release of males only or both sexes should be attempted.