An experimental investigation of the reliability and feasibility of a regenerative cooling system in hybrid rocket engines based on saturated nitrous oxide is presented. The novelty of the work consists in the use of a cooling system based on saturated liquid nitrous oxide, wrapping the external surface of a COTS graphite nozzle. The detrimental heat fluxes developed at the throat are handled with a slight increase in assembly complexity. Ten firing tests are performed with an incremental logic of the mass flow rate from 9 to 45 g/s, which is controlled by cavitating the fluid through orifices with different diameters upstream of the cooling system. The tests have been performed at different ambient temperatures in different seasons. The multiphase state of the coolant is evaluated by measuring pressure and temperature upstream and downstream of the cavitation orifice and cooling channels. The test with the smallest mass flow rate developed the highest vapour fraction downstream of the orifice and cooling channels at 0.458 and 0.481, respectively. The cooling performance is indirectly evaluated by measuring the nozzle temperature at 3, 5, and 8.5 mm from the throat internal surface. The results show that steady temperatures are achieved inside the nozzle, with throat temperatures included between 700 and 1200 K in a chamber pressure range between 5 and 30 bar. Nozzle erosion never occurs in the entire experimental campaign, and the nozzles are totally reusable for more ignitions. The coolant heat transfer coefficient increased from 3912 to 21181 W/(m2∙K) by increasing the flow rate per channel from around 3 to 15 g/s. Compared to cryogenic oxygen, liquid nitrous oxide displays higher cooling performance. Finally, the cavitation orifice is moved downstream of the cooling system, showing worse performance of the overall propulsion system.