Hydrogen peroxide (HOOH) in ice and snow is an important chemical tracer for the oxidative capacity of past atmospheres. A previous study on the photolysis of HOOH used ice samples made from solutions frozen over the course of minutes, conditions that should result in HOOH present in quasi-liquid layers (QLLs). However, since HOOH in natural snows and ices can also be present in the ice matrix, in this work we studied HOOH photolysis in ice samples prepared by flash freezing very dilute aqueous solutions, to keep HOOH in the ice matrix. In addition, rather than following HOOH photolysis by using an organic probe to scavenge photoformed OH, here we measured changes in HOOH concentrations. Based on initial rates of HOOH photolysis, our resulting quantum yields for OH formation, Φ(OH), in ice range from approximately 0.25 to 0.43 between 238 and 265 K, respectively. These values are up to 50% smaller than previously published values from the QLL HOOH experiments, indicating that HOOH in the ice matrix undergoes slower photolysis. Further, during our experiments the HOOH photolysis rate constant sometimes shows a biphasic behavior, with slower photolysis after several hours of illumination. Using these final rate constants of HOOH decay we derive a lower bound, temperature-independent value for Φ(OH) of 0.29(±0.10). Based on Φ(OH) values from our initial rate measurements, we calculate HOOH lifetimes in polar environments as large as 30 days at the summer solstice, which is too short to allow burial of HOOH beneath the photic zone in the snowpack.