This work contributes to the understanding of physical mechanisms that control flashback, or more appropriately combustion recession, in diesel sprays. A large dataset, comprising many fuels, injection pressures, ambient temperatures, ambient oxygen concentrations, ambient densities, and nozzle diameters is used to explore experimental trends for the behavior of combustion recession. Then, a reduced-order model, capable of modeling non-reacting and reacting conditions, is used to help interpret the experimental trends. Finally, the reduced-order model is used to predict how a controlled ramp-down rate-of-injection can enhance the likelihood of combustion recession for conditions that would not normally exhibit combustion recession.In general, fuel, ambient conditions, and the end-of-injection transient determine the success or failure of combustion recession. The likelihood of combustion recession increases for higher ambient temperatures and oxygen concentrations, as well as for higher reactivity fuels. The likelihood of combustion recession was further linked to the characteristics of end-of-injection entrainment. A simple equation, linking equivalence ratio at the lift-off length, ϕ(LOL), with a dimensionless parameter related to the end-of-injection entrainment wave, was found to well predict the propensity for combustion recession of dodecane sprays over a wide range of experimental data and 1-D model predictions. Our results suggest that this relationship is ϕ(LOL) ∼ Deff/(αUeff)0.474, where Deff is the effective orifice diameter, αα is the end-of-injection ramp-down duration, and Ueff is the effective injection velocity.