The US Federal Aviation Administration (FAA) observed during explosion tests that at a low concentration of agent, some candidate halon replacement agents increased the explosion severity instead of mitigating the event. At UTC Aerospace Systems a test program was developed to assess the behaviour of alternative agents at values below inerting concentration. Two agents were selected, C2HF5 (Pentafluoroethane, HFC-125) and C6F(12)O (FK-5-1-12, Novec (TM) 1230). Baseline tests were performed with unsuppressed C2HF5 (propane)/air mixtures and C3H8/air mixtures with CF3Br (Halon 1301) and N-2 (nitrogen). Using CF3Br or N-2 at below inerting concentrations mitigated the explosion. C2HF5 was tested against C3H8 at stoichiometric (4 vol%) and lower explosion limit (LEL) (2 vol%). Against 4 vol% C3H8 the combustion was mitigated, proportional to agent concentration; however, low concentrations of C2HF5 with 2 vol% C3H8 enhanced the explosion. Tests with N-2 against a volatile mixture of C3H8 with C2HF5 showed that N-2 mitigated the events. Final tests were performed with low concentrations of C8F12O against C3H8/air mixtures. This showed similar behaviour to that observed with the C2HF5 tests. Normally during qualification tests for new agents the stoichiometric concentration of a fuel is deemed to be the worst case scenario and the baseline against which agents are tested. The above described test results show that this assumption may need to be reconsidered. This work shows that contrary to common assumption the agents investigated are unlikely to have acted chemically at the flame front, but most likely, mainly cooled the flame and changed the stoichiometry, i.e. the ratio of components of the flammable mixture. (C) 2015 Elsevier Ltd. All rights reserved.