Abstract : This paper presents a constrained Tucker-3 model for blind beamforming. The constrained structure of this model is parameterized by two constraint matrices that capture the spatial and temporal structure of the channel. We first present a generalized Tucker-3 model that considers a multipath propagation scenario with large delay spread and different number of paths per source. This tensor model generalizes the models of [N.D. Sidiropoulos, G.Z. Dimic, Blind multiuser detection in WCDMA systems with large delay spread, IEEE Signal Process. Lett. 8 (3) (2001) 87-89] and [A. de Baynast, L. De Lathauwer, Détection autodidacte pour des systèmes à accès multiple basée sur l'analyse PARAFAC, in: Proceedings of XIX GRETSI Symposium on Signal and Image Processing, Paris, France, 2003] by considering a more general multipath propagation scenario. From this general model, two particular cases can be derived by adjusting the constrained structure of the Tucker-3 core tensor. Identifiability of the proposed tensor model is studied and some simulation results are presented to illustrate the application of this model to blind beamforming.