A　novel　method　for　estimating　2D　direction　of　arrival　(DOA)　of　broadband　signals　is　proposed.　It　is　called　two-dimensional　generalized　spectral　correlation　signal　subspace　fitting　or　2D　GSC-SSF.　Exploiting　the　property　that　the　cyclic　correlation　operation　transfers　the　time　delay　of　the　signal　to　phase　shift　of　the　cyclic　autocorrelation　function,　the　method　combines　the　features　of　the　GSC-SSF　and　DOA　matrix　method　and　is　capable　of　estimating　2D　DOA　of　both　narrowband　and　broadband　signals　effectively.　The　theoretical　analysis　and　simulation　results　show　that　the　method　inherits　the　advantages　of　cyclic　DOA　estimation　methods　in　signal　selectivity,　accuracy,　resolution,　overload　ability　and　detecting　ability,　and　also　possesses　the　same　features　as　the　DOA　matrix　method,　i.e.,　no　2D　search　or　parameter　alignment　is　needed.