Remote observation using light detection and ranging (lidar) offers a unique opportunity to quantify the geometry and size structure of tropical forests contiguously at fine spatial resolution. We used ten samples of airborne lidar data from tropical forests – six regions from Brazil, and other Neotropical forests such as Peru, French Gui, Panama and Costa Rica. We quantify gap size frequency distribution along vertical and horizontal dimensions in ten Neotropical forest canopies distributed across gradients of climate and landscapes using airborne lidar measurements. We found that natural forest disturbances (tree-fall gaps) follow a power-law distribution. Mean gap area (50 to 900 m2) and frequency (10,960 to 27,158) varied considerably among sites (200-ha each). However, we found that imposing a minimum gap area (20 m2) constrained the exponent of the power-law fit of gap frequency to a narrow range from -1.2 to -1.3. This contrasts with previous studies that included smaller canopy gaps (1-20 m2). The convergence of gap frequency distribution represented by the narrow range of the power-law exponents found in this study suggests an invariant scaling property of gaps in Neotropical forests.