Most previously proposed mechanisms for the pyrolysis of diborane rest on the assumed symmetric dissociation B 2H 6 ⇌ 2BH 3, which as an equilibrium has never been observed. On exhaustive re-examination the existing evidence strongly implies that (under ordinary conditions) the initial act of decomposition is neither symmetric nor asymmetric fission, but 2B 2H 6 → BH 3 + B 3H 9, followed so rapidly by BH 3 + B 2H 6 → B 3H 9 that a measurable concentration of BH 3 never builds up. The straight pyrolyses of B 2H 6, B 4H 10 and B 5H 11, the effect of H 2 on the decomposition of each compound, the B 2H 6B 4H 10, B 2H 6B 5H 11 and B 2H 6B 5H 9 co-pyrolyses, the interconversion of the boranes and their behaviour in D and 10B isotope-exchange studies are all explained on the basis of a single unified mechanism as far as the appearance of B 10H 14. The paradox arising from an apparent conflict in the evidence regarding the first isolable intermediate in the pyrolysis of diborane, whether B 4H 10 or B 5H 11, is resolved in favour of the former. Interconversion reactions occur by the exchange of BH or BH 3 (but never BH 2) groups, or, in the case of hydrogen-rich species of formula type (BH) n (BH 3) 3, by splitting off BH 3 molecules. Intermediates participate in condensation reactions in which normally H 2 is released, but B 3H 9 appears to be the only species involved that splits off H 2 spontaneously at moderate temperatures.