Two experimental studies have been made of the gas phase oxidation of diborane. First, conventional oxidation in furnace-heated vessels, at temperatures between 80 and 200 °C was used to show that degenerately branching chains must be involved. Secondly, some transient intermediates were monitored by kinetic absorption spectroscopy in oxidations initiated by flash photolysis. Since diborane absorbs only weakly above 2000 A, only very pure silica (Spectrosil A) would transmit enough energy to produce dissociation. The spectrum of BH and a pressure change were evidence of this when diborane was flashed alone at about 0*01 atm. A very rapid reaction was produced on flashing diborane with oxygen both at low total pressures, about 0-03 atm, and with the total pressure raised to about 1 atm by the addition of inert gas. In apparatus of conventional silica, a chromophoric sensitizer (nitrogen dioxide) was necessary. An incubation period of about 100 us followed the flash, towards the end of which the spectra of OH and B 02 appeared. With low total pressure, BO and BH also appeared as the temperature began to rise in the developing explosion. The gaseous products from both types of reaction were analysed, and a reaction scheme is proposed which takes account of them and of the observed radicals. The hydroxyl radical is undoubtedly intimately concerned in the principal reactions. Predictions made from this scheme for non-explosive conditions are in fair agreement with observed kinetics of the slow oxidation.