We propose a blind direction of arrival (DOA) estimation method for multicarrier code-division multiple access (MC-CDMA) systems, which does not require knowledge of the spreading codes of transmitters. Our method involves reconstructing a fourth-order tensor with observations from the antenna array at the receiver. We then apply higher-order singular value decomposition (HOSVD) to separate the signal subspace from the noise subspace. To enhance the sparsity of the solution, we develop a smoothed $\ell _{0}$ -norm sparse reconstruction method by designing continuous functions with reweighted vectors. Finally, DOA estimates are obtained by searching the spectrum of the sparse solution using an iterative method that ensures fast convergence. Numerical simulations show its superiority over previous work.