Polarization tests were conducted on proton exchange membrane fuel cells (PEMFCs) at cell temperatures between 60 and 80degreesC with various reactant humidification levels; varied from no external humidification to fully saturated on both the anode and the cathode. Elimination of cathode external humidification, while maintaining a fully humidified anode inlet, resulted in cell performance loss of only 5% or 33 mV (from 0.674 to 0.641 V) at 400 mA/cm(2) at an anode stoichiometry of 3 and a cathode stoichiometry of 4. When both the anode and the cathode humidification were removed ("dry operation"), cell performance strongly depended on the cell operating temperature and the inlet gas stoichiometric flow rates. High performance of non-humidified PEMFCs was demonstrated by optimizing the operating cell temperature and the inlet gas stoichiometric flow rates, to find a balance between cell "flooding", oxygen mole fraction, and proton conductivity. Performance of a cell with an in-house cathode gas diffusion layer (GDL) showed a loss of only 4% or 29 mV (from 0.674 to 0.645 V) compared to the near-saturated condition, at 400 mA/cm2 and optimum stoichiometry. Using a commercial E-TEK-V.2.11 carbon cloth cathode GDL showed significantly greater loss when operated with no external humidification at the same optimized condition. An overall system analysis suggested that at optimum stoichiometric flow rates and cell temperatures, with the in-house GDL, while the net power output might be reduced by at most 17% under dry operation the total required non-stack energy duty would be cut by over 46% when compared to operation with saturated inlet gases. (C) 2004 Elsevier B.V. All rights reserved.