We demonstrate that two recent innovations in the field of practical quantum key distribution (one-way autocompensation and passive detection) are closely related to the methods developed to protect quantum computations from decoherence. We present a new scheme that combines these advantages, and propose a practical implementation of this scheme that is feasible using existing technology.PACS numbers: 03.65. Ud, 03.67.Dd, 03.67.Lx, 42.65.Ky Decoherence has been a principal impediment in quantum information processing applications. In quantum computing, decoherence-induced deviations from the desired computational trajectory at the single-qubit level will quickly accumulate if left uncorrected. Thus, techniques such as decoherence-free subspaces (DFSs, for a review, see Ref.[1]) have been developed as tools for protecting quantum computations. In quantum key distribution (QKD, for a review, see Ref.[2]), single-qubit errors are also deleterious; however, sufficiently infrequent single-qubit errors are tolerable, since the resulting errors can be corrected by classical error correction protocols. This has led many QKD experimentalists to forego the complexity of decoherence-mitigation techniques such as DFSs in favor of more conventional methods to improve the precision of single-qubit operations (periodic alignment of polarization axes, temperature stabilization of interferometers, etc.). In this letter, we consider the applicability of DFSs to QKD. This letter is organized as follows. We begin by demonstrating that a recently-proposed QKD implementation (one-way autocompensating quantum cryptography [3]) is, in fact, equivalent to a well-known DFS. We then pursue a suggestion in Ref.[2] to consider a single-qubit, phase-time coding QKD scheme in which Bob is not required to actively switch between conjugate measurement bases. We show that both one-way autocompensation (OWA) and passive detection are achieved by embedding the logical Hilbert space in a larger physical Hilbert space. Next, we describe a new scheme that combines OWA and passive detection. Finally, we propose an experimental implementation of this new scheme that is feasible using existing technology.Relating OWA and DFSs.-In Ref.[3], Klyshko's "advanced wave interpretation" [4] was used to describe OWA as a variation on round-trip autocompensation [5,6]. These schemes are called autocompensating because they allows high-visibility quantum interference without calibration or active stabilization of the receiver's (Bob's) apparatus. In the context of quantum computation theory [7], a more natural explanation of OWA is provided by DFSs. Palma et al. [8] have shown that a single logical qubit encoded in two physical qubits according to( 1) will be protected against collective dephasing. Collective dephasing describes a noise model in which each physical qubit is subject to the same transformationwhere φ is an uncontrolled degree of freedom. Under this transformation, the states |01 and |10 acquire the same phase factor (e iφ ). Thus, a qubit encode...