In response to observed inoculation failures in the tropical forage legumes Desmanthus spp. and Stylosanthes seabrana, research was initiated to improve the survival of rhizobia on seed at high temperature. We compared the survival of rhizobia in freeze-dried and conventional peat inoculants in the laboratory at 30–50°C and 5–8% relative humidity. Higher numbers of rhizobia [>10⁶ colony forming units (cfu)/seed] were counted on seed inoculated with freeze-dried rhizobia compared with seed inoculated with peat (≤10⁴) in all treatments at all sampling times (0, 2, 7, 14 and 21 days after inoculation). Increasing the incubation temperature to 70°C significantly (P<0.05) reduced the number of freeze-dried rhizobia surviving on seed, but rhizobia were detectable on Desmanthus seed 1, 2 and 7 days after inoculation and on Stylosanthes seed 1 and 2 days after inoculation. Freeze-dried rhizobia stored over silica gel in the laboratory for 325–349 days lost 3.5–4.5 log₁₀ cfu/g. Freeze-dried rhizobia inoculated on seed and stored in a farm shed from April to July lost 0.6–0.8 log₁₀ cfu/seed.month. Survival of rhizobia in peat and freeze-dried inoculants on seed over 1–4 weeks in a glasshouse without air conditioning (18–51°C) was poor (0–1.2 log₁₀ cfu/seed at all sampling times), and was attributed to exposure to a combination of high temperature and ambient relative humidity. Laboratory studies confirmed that increasing relative humidity from 5–8 to 31–63% reduced the survival of rhizobia in freeze-dried inoculants on seed, particularly at 50°C. Further work is required to improve the survival of freeze-dried inoculants at a relative humidity greater than 5–8% and under long-term storage. Optimisation of this technology has the potential to increase inoculation success for all legumes sown at high temperature in Australia. Freeze-dried inoculant technology may also have an application in seed preinoculation.