The present invention relates to a method for calibrating a system (10) comprising an external eye-tracking device (16) and a computing device (12a; 12b) and for capturing the gaze of a user (P) on the screen (13) of the computing device in real-time. The calibration of the system (10) comprises: capturing with one or more cameras (17, 18) of the eye-tracking device (16) at least of one image of landmarks (f1, f2, f3...fn) of the face of the user (P) to identify the 3D position of each landmark in the coordinate system (ECS) of said eye tracking device; capturing with a camera (15) of the computing device (12a; 12b) the same landmarks (F1, F2, F3) of the face of the user (P) in the image coordinate system (ICS) of the computing device camera (15) to identify the 2D position of each landmark in the image coordinate system ICS; computing the 3D pose of the camera (15) of the computing device (12a, 12b), defined as the camera coordinate system (CCS), as a function of the 3D position and 2D position of each landmark (f1, f2, f3...fn) respectively in the coordinate system ECS and in the coordinate system ICS, and computing the 3D pose of the screen of the computing device, defined as the screen coordinate system (SCS), as a function of the camera coordinate system and mechanical parameters describing how the screen (13) is positioned with respect to the camera (15) of the computing device. Capturing the gaze of a user (P) on the screen (13) of the computing device in real-time comprises: retrieving a gaze ray (d) of the user (P) with the eye-tracking device (16), and intersecting the gaze ray (d) of the user (P) with the plane of the screen of the computing device, as a function of the ECS and SCS parameters, to capture the gaze-on-screen in real-time.