An order recursive algorithm is proposed to solve the normal equation in estimating continuous-time AR process parameters. The new algorithm requires O(M 2 ) computations, where M is the order of the AR process, and therefore is computationally efficient. This is similar to the Levinson-type algorithm of Pham and Le Breton. However, the proposed algorithm has a new associated lattice structure, and as a consequence a new stability test for continuous-time systems, not just AR processes, by checking the reflection coefficients of the lattice. This can be viewed as a continuous-time analogy that is recursive-in-order of the well-known discrete-time version. Its discrete-time approximation, when sampled-data are given, and connections with other algorithms are also discussed.