The high powerfulness of TLS technique for quick 3D data acquisition is extending its use to many fields. To further reduce the surveying time and to simplify all operational tasks, the TLS direct georeferencing may be a very suitable approach instead of the technique based on ground control points (targets). This chance is allowed by the most part of existing instruments, as a default or as an optional capability. The paper describes the geometric model involved in the direct georeferencing, considering scanners mounted either in vertical and in tilted position. Secondly, an analysis of errors affecting laser scanners measurement is proposed. The total error budget results from the propagation of errors due to intrinsic measurements and to the adopted georeferencing technique. Here errors connected to the instrument setup needed to get direct georeferencing are analized. Finally, a simulation finalized to define the achievable accuracy in 3D point measurement according to different sets of instrumental parameters is proposed. Furthermore, simulated data have been compared to a real case of data acquisition performed by means of both direct georeferencing and by the use of ground control points.