Possible pathways for translocation across the membrane in active transport processes are examined theoretically. Thermodynamic and kinetic requirements are readily satisfied by an alternating-access mechanism of the kind that has been proposed in the past by several investigators. The essential features of this mechanism (for transport of a single species) are shown to be defined by four explicit conditions. (i) The transport protein must have at least two distinct conformational states, each accessible from only one side of the membrane. (ii) Binding affinity for the transported species is high in the state accessible from the uptake side of the membrane and much lower in the state accessible from the discharge side. (iii) The change from one conformation to the other involves movement of the binding site itself (with the transported species remaining attached) or rearrangement within the site that is topologically equivalent to such movement. (iv) Return to the original conformation occurs with unoccupied binding sites. The analysis demonstrates that a passage through the membrane that is simultaneously accessible from both sides cannot be used for active transport regardless of what the energetics of opening or closing of the passage may be. Even movement from one fixed site to another within the protein, without access to the outside, is virtually excluded as a possible element of the central mechanism. A ligand conduction mechanism for ATP-linked ion transport is in principle conceivable but is subject to restrictions that make it improbable.The device used by transport proteins to allow a transported species to traverse an otherwise essentially impermeable membrane must be different in active and passive transport systems. In active transport, translocation is thermodynamically uphill, requiring some kind of boost from the protein. In passive transport, translocation is thermodynamically downhill, and nothing more is needed than a simple pathway across the membrane, simultaneously accessible from both sides. Many investigators have recognized this difference, and the most frequently proposed model for active transport involves two conformational states of the protein, with alternate access from the two sides of the membrane, permitting the requisite boost to the transported species to take place in the transition from one conformational state to the other (1-8). However, some investigators still view the translocation domain of an active transport protein as resembling a channel or pore of the type that is used to facilitate passive transport (9, 10). Yet another proposal is provided by the ligand conduction model of Mitchell (11) in which the uphill boost to the transported species is provided by direct chemical interaction with the free energy-donor species, instead of through the protein. The existence of these contrary views has prompted the theoretical investigation described in this paper.Relative contributions of osmotic and electrical components to the work of transport are not relevant t...