We show that quantum feedback control can be used as a quantum error correction process for errors induced by weak continuous measurement. In particular, when the error model is restricted to one, perfectly measured, error channel per physical qubit, quantum feedback can act to perfectly protect a stabilizer codespace. Using the stabilizer formalism we derive an explicit scheme, involving feedback and an additional constant Hamiltonian, to protect an ($n-1$)-qubit logical state encoded in $n$ physical qubits. This works for both Poisson (jump) and white-noise (diffusion) measurement processes. In addition, universal quantum computation is possible in this scheme. As an example, we show that detected-spontaneous emission error correction with a driving Hamiltonian can greatly reduce the amount of redundancy required to protect a state from that which has been previously postulated [e.g., Alber \emph{et al.}, Phys. Rev. Lett. 86, 4402 (2001)].

11 pages, 1 figure; minor corrections