During the summer the Arabian Gulf is the world's warmest sea, also characterized by hypersalinity and extreme annual temperature fluctuations (12–35oC), making it marginal for coral growth. Yet extensive reefs occur in all eight nations bordering the Gulf. Here we present data demonstrating recurrent summer hypoxia events [oxygen concentration (O2) <2 mg l−1] at a reef in the southern Gulf. Currently these episodes are short enough (median 3 h, max 10 h) to preclude mass mortality. Will this always be the case? Predicting future Gulf hypoxia risk for coral reef ecosystems requires diagnosing the underlying causes driving the timing and magnitude of O2 swings. To this end, we compare our data with the output of a simple coupled 1-D water column/biogeochemical model of the reef environment. This allows us to give quantitative estimates of the O2 fluxes produced by photosynthesis both in the water column and within the coral framework, by respiration processes in the benthos, and from the atmosphere. We demonstrate the role of turbulent mixing, and in particular of tides, in shaping the temporal variability of the amplitude of the diel O2 cycle. We find that, in spite of significant turbulent mixing, which maintains the temperature vertically well-mixed, the biological O2 production and consumption is dominant over the atmospheric O2 flux, and is sufficient to generate vertical differences of 1 to 5 mg l−1 between the bottom and 1.5 m above it. While estimating future trends of hypoxia frequency will require further study, the present findings single out the relevant physical and biological processes (and their interplay) which deserve further scrutiny. The Gulf today experiences temperatures expected to occur across much of the tropics by the end of the century, and the observation of recurrent hypoxia events in the Gulf suggests that similar hypoxic phenomena may represent an important, but to date underappreciated, threat to the future of global coral reefs.