For reducing costs of market frictions, investors need to build a small-scale portfolio by solving a cardinality constrained portfolio selection problem which is NP-hard in general and not easy to be solved eciently for a large-scale problem. In this paper, we propose a penalty proximal alternat- ing linearized minimization method for the large-scale problems in which a sequence of penalty subproblems are solved by utilizing proximal alternating linearized minimization frame and sparse projection techniques. For exploit- ing the structure of the problems and reducing the computation complexity, each penalty subproblem is solved by alternately solving two projection sub- problems. The global convergence of the method to a KKT point or a local minimizer of the problem can be proved under the characteristic of the prob- lem. The computational results with practical problems demonstrate that our method can nd the suboptimal solutions of the problems eciently and is competitive with some other local solution methods.