Abstract: In order to solve the off-grid representation of the steering vector that exist in the existing sparse reconstruction algorithms for DOA estiamtion,an approach for DOA estiamtion of cyclostationary signals based on continuous sparse recovery algorithm is proposed in this paper. Firstly，the conventional spectral correlation signal subspace fitting algorithm is analyzed and researched．Then by using the continuous sparse recovery in the circulation domain to construct sparse recovery model of the cyclostationarity signal. Compared with the conventional Cyclic MUSIC algoritthm and the existing sparse reconstruction algorithms，the proposed method can overcome the off-grid problem,has good sparse recovery precision and better performance of sparse recovery, and is also suitable to the scenario that the number of signals is more than that of array elements. Theoretical analysis and simulation results prove the validity of the proposed algorithm.