In the process of tunnel construction and operation, the traditional method of crack inspection in tunnel operation is through manual photographing and field recording which is not only labor-intensive with lower data quality but also hard to obtain the important geometric information of the cracks. With the development of terrestrial laser scanning technology, Three-dimensional point cloud collection from a tunnel becomes feasible and affordable. In this paper, we propose an automatic crack detection method to automatically identify the cracks from the 3D point cloud of a tunnel and also extract the corresponding dimension information of the cracks. Through a field experiment with a tunnel project, the feasibility of the proposed method was validated, and the experiment results showed that the proposed method could detect the severe crack diseases with a width of more than 0.6 mm with the relative errors of the width and the length within 10%.