Unlike N fertiliser, inoculation of crops with diazotrophic bacteria does not reliably increase crop yield. Inoculants added to plants are often ineffective due to an inappropriate mechanism of action that does not add substantial nitrogen to the system, low viability of the inoculant or poor persistence of the bacteria in the soil. In this paper we identify the major problems of using inoculants, and dicuss how improved effectiveness may be achieved by selecting strains of bacteria capable of tolerating soil stresses while effectively fixing nitrogen. The key elements of an effective inoculant are identified, as a mechanism that increases plant available nitrogen, a high quality inoculant appropriate to the crop cultivar being grown and an inoculant which is competitive and robust enough to tolerate the soil conditions it will encounter. We discuss the advantages of using specific N2 fixing bacteria isolated from stressful soil environments as inoculants and illustrate, with examples from the literature, how such bacteria have been shown to provide effective inoculants for specific crops which are subsequently cultivated in those soils.