1 The sterile insect technique (SIT) involves the release of large numbers of sterile or partially-sterile insects into a wild pest population to dilute the number of successful wild matings, with the eventual aim of eradication or area-wide suppression. General population models, encompassing a wide range of SIT types, were used to derive principles for optimizing the success of SIT, with particular emphasis on the application of partial sterility leading to inherited sterility in the F 1 population. 2 The models show that inherited sterility can only be guaranteed to be more effective than complete sterility if matings between irradiated-lineage partners are unsuccessful. This is widely assumed but rarely examined experimentally. 3 The models allow the critical overflooding ratio, c , to be calculated for a particular target species, suggesting the release rate required to prevent population increase. Successful eradication using SIT alone should aim for a substantially higher release rate than suggested by c . 4 The models show that pest populations may continue to increase in the first few generations of SIT releases, regardless of release rate, as irradiated-lineage individuals infiltrate the population. This does not necessarily imply that the SIT programme will be unsuccessful in the longer term. 5 For pests with overlapping generations, the models suggest that frequent small releases may be more effective than less frequent large releases, particularly when the average release rate is close to the critical threshold for success.