One of the challenges in real-time monitoring of airborne alpha-emitting radionuclides using Continuous Air Monitors (CAMs) is to discriminate alpha particles emitted by artificial radionuclides from those emitted by natural radionuclides, as the alpha energy spectrum is easily degraded due to its high stopping power. Although the measurement geometry significantly affects the alpha energy spectrum, there are no reports that have investigated the optimized geometry. This study conducted a Monte Carlo simulation to optimize the measurement geometry of the CAMs. The CAM was modelled using a radiation transport simulation code, and the counting efficiencies for 218 Po and 239 Pu and the overlapping ratio were simulated. The Minimum Detectable Concentration (MDC) of 239 Pu was then estimated as an indicator to optimize the geometry. Although a constant counting efficiency is generally used for the same geometry regardless of the alpha energy, the counting efficiency differed up to ∼0.04 by the alpha energy depending on the measurement geometry. The overlapping ratio for a 50-mm diameter detector ranged from 0.22–0.35, while those for a 25-mm diameter detector was below 0.25. This discrepancy is because of the incident angle distribution biased towards large angles. The lowest MDC was estimated to be ∼3 Bq m -3 at 20 L min -1 under 25 Bq m -3 of 218 Po. The MDC was low for small filter diameters and Filter-to-Detector Distances (FDDs) under most of the simulated conditions. The results indicate that a shorter filter diameter and FDD would be preferable for alpha spectrometry under normal air pressure. • Measurement geometry of continuous air monitors was optimized. • Counting efficiency differed by the alpha energy depending on the geometry. • Biased incident angle distribution resulted in high overlapping ratio. • Short filter diameters and filter-to-detector distances would be preferred.