Ethanol is a major confounder in gas sensing because of its omnipresence in indoor air and breath from disinfectants or alcoholic beverages. In fact, most modern gas sensors (e.g., graphene, carbon nanotubes, or metal oxides) are sensitive to ethanol. This is challenging because ethanol is often present at higher concentrations than target analytes. Here, a simple and modular packed bed filter is presented that selectively and continuously removes ethanol (and other alcohols like 1-butanol, isopropanol, and methanol) over critical acetone, CH4, H2, toluene, and benzene at 30-90% relative humidity. This filter consists of catalytically active ZnO nanoparticles (dBET = 55 nm) made by flame aerosol technology and annealing. Continuous oxidation of ethanol to CO2 and H2 was observed at filter temperatures above 260 °C while below that, unwanted acetaldehyde was formed. Most remarkably, ethanol concentrations up to 185 ppm were removed from exhaled breath in preliminary tests with an alcohol intoxicated volunteer, as confirmed by mass spectrometry. At the same time, almost 4 orders of magnitude lower (e.g., 0.025 ppm) acetone concentrations were preserved. This was superior to previous catalyst filters (e.g., CuO, SnO2, and Fe2O3) with overlapping ethanol and acetone conversions and related to ZnO's surface basicity. The ZnO filter performance was stable (±2.5% conversion variability) for, at least, 21 days. Finally, when combined with a Si-doped WO3 sensor, the filter effectively mitigated ethanol interference when sensing acetone without compromising the sensor's fast response and recovery times. Such catalytic filters can be combined readily with all gas sensors.