We have evaluated the aerosol passing efficiency of a variety of inlet systems during three experiments on the National Center for Atmospheric Research Electra. During the Dynamics and Chemistry of the Marine Stratocumulus (DYCOMS) program, we found discrepancies between concentrations in cloud water and the air below cloud, which we attribute to curved-inlet aerosol losses. In the First ISCCP Regional Experiment (FIRE) program, the same curved inlet passed significantly less material than a straight one. In the Particulate Matter Airborne Sampling Inlet Experiment (PASIN), we analyzed the material deposited within one inlet tube to establish an efficiency reference. Six different inlets were used simultaneously with six filter samplers to evaluate the effects of tube diameter, radius of curvature, and surface coating. The PASIN filters collected different amounts of material when they should have been sampling the same aerosol. Extraction of the 1-inch metal reference tube routinely found that 50--90% of the aerosol material had been deposited in the tube. None of the inlets had an average bulk efficiency greater than about 50% for the aerosol species we measured. In the commonly used 1-inch ID curved tube inlet, only 10-20% of marine sodium made it to the filter. High turbulence in the diffuser cone may be responsible for much of the observed loss. It is possible that many of the existing aircraftderived aerosol data in the literature are serious underestimates of actual ambient concentrations. We suggest some changes in strategy to improve the collection of aerosols from aircraft.