Development of laser scanning has offered great opportunity to capture three-dimensional (3D) topographic information of objects’ surfaces in a highly accurately and efficiently. Particularly, a terrestrial laser scanner is able to acquire millions of points within a second at millimetre accuracy. This technology has been widely used in many civil engineering applications, including surveying, construction management, and infrastructure inspection. Traditionally, tank inspection was carried out on-site by physical inspectors with suitable measurement equipment (e.g. tapes, staffs and a total station). This approach, albeit the most common one, has many downsides: subjective results, slow and expensive procedure, requirement of experienced and trained inspectors and close service of the tank. Additionally, all results are stored as hard copies, which lead to difficulties in tracking damage development and management. To mitigate these disadvantages, this paper proposes a method for (i) automatically extracting a point cloud of a tank wall from a massive data points, and (ii) evaluating the tank wall through its deformation.