Frost formation in air-to-air heat/energy exchangers is a challenging problem in regions with very cold weather conditions, when energy recovery is most needed. Membrane-based energy exchangers may assist in overcoming frosting. In this paper, an experimental facility was developed to enable researchers to test air-to-air exchangers under frosting conditions. The test setup, sensors and data acquisition system, uncertainty bounds for major parameters, and experimental procedure are described. Two geometrically identical plate exchangers: one with a water vapor permeable membrane (energy exchanger), and one with an impermeable plate (heat exchanger) were tested under frosting conditions. The first step in frosting experiments is to find a reliable and cost-efficient method to detect frosting in the exchangers. Four methods were used to determine the onset of frost growth: visual inspection, change in the effectiveness (Δε), change in the pressure drop across the exchanger (Δp), and change in the outlet temperature (ΔT). The main contributions of the paper are the detailed comparison of these methods and the introduction of a new frost detection method based on temperature measurements only (ΔT method).It is concluded that the Δp and ΔT methods were more reliable and practical than the other methods for both heat and energy exchangers. The ΔT method detected frosting sooner and with lower uncertainties than the other methods. Furthermore, the ΔT method was least affected by the operating conditions. On the other hand, the ΔP method gave a better indication of the severity of frosting in the exchanger as more frost resulted in a higher pressure drop across the exchanger.