The generally accepted mechanism leading to ozone depletion events in the Arctic assumes efficient gas-phase recycling of bromine atoms, such that the rate of ozone depletion has often been estimated as the rate that Br atoms regenerate through gas-phase BrO + BrO and BrO + ClO reactions. Using a large suite of data from the OASIS2009 campaign, our modeling results show that the gas-phase regeneration of Br is less efficient than expected and that heterogeneous recycling on surfaces is critical.