This work explores, for the first time, the application of a Terrestrial Laser Scanner (TLS) and a comparison of point clouds in the 4D monitoring of active sinkholes. The approach is tested in three highly-active sinkholes related to the dissolution of salt-bearing evaporites overlain by unconsolidated alluvium. The sinkholes are located in urbanized areas and have caused severe damage to critical infrastructure (flood-control dike, a major highway). The 3D displacement models derived from the comparison of point clouds with exceptionally high spatial resolution allow complex spatial and temporal subsidence patterns within one of the sinkholes to be resolved. Detected changes in the subsidence activity (e.g., sinkhole expansion, translation of the maximum subsidence zone, development of incipient secondary collapses) are related to potential controlling factors such as floods, water table changes or remedial measures. In contrast, with detailed mapping and high-precision leveling, the displacement models, covering a relatively short time span of around 6 months, do not capture the subtle subsidence (<0.6–1 cm) that affects the marginal zones of the sinkholes, precluding precise mapping of the edges of the subsidence areas. However, the performance of TLS can be adversely affected by some methodological limitations and local conditions: (1) limited accuracy in large investigation areas that require the acquisition of a high number of scans, increasing the registration error; (2) surface changes unrelated to sinkhole activity (e.g., vegetation, loose material); (3) traffic-related vibrations and wind blast that affect the stability of the scanner.