Stress fractures have been proposed to arise from repetitive activity of training inducing an accumulation of microfractures in locations of peak strain. However, stress fractures most often occur long before accumulation of material damage could occur; they occur in cortical locations of low, not high, strain; and intracortical osteopenia precedes any evidence of micro-cracks. We propose that this lesion arises from a focal remodeling response to site-specific changes in bone perfusion during redundant axial loading of appendicular bones. Intramedullary pressures significantly exceeding peak arterial pressure are generated by strenuous exercise and, if the exercise is maintained, the bone tissue can suffer from ischemia caused by reduced blood flow into the medullary canal and hence to the inner two-thirds of the cortex. Site specificity is caused by the lack, in certain regions of the cortex, of compensating matrix-consolidation-driven fluid flow which brings nutrients from the periosteal surface to portions of the cortex. Upon cessation of the exercise, re-flow of fresh blood into the vasculature leads to reperfusion injury, causing an extended no-flow or reduced flow to that portion of the bone most strongly denied perfusion during the exercise. This leads to a cell-stress-initiated remodeling which ultimately weakens the bone, predisposing it to fracture.