State of the art graph-based 3D simultaneous localization and mapping (SLAM) systems are typically divided into the so-called front- and back end. The front end’s task is to align laser scans locally, constructing a graph of the measurements in the process. The back end’s task is to optimize this graph in order to find a maximally consistent configuration of the nodes. The well-known iterative closest point (ICP) algorithm is often used to align pairs of laser scans. As it only provides local convergence, it generally requires an approximate guess in order to determine the correct relative transformation. In practice, this guess is often obtained from a robot’s odometry. However, in some applications this odometry information is not available and a fast initial guess is needed for real-time operation. Existing feature-based approaches are often unsuitable for real-time operation due to their computational requirements. The authors present a featureless algorithm that is able to compute an approximate transformation between two laser scans quickly, serving as initial guess for ICP. This algorithm is an adaptation to 3D of a 2D correlative scan-matching algorithm by Olson and will be evaluated in the context of 3D SLAM. Comparisons are made to results that use features as they are used in image processing to match images. The authors make some plausible and moderate assumptions on the scan acquisition and experimentally show that these are usually met. The experiments show that the presented algorithm is able to align two 3D laser scans quickly and reliably in the context of mobile robotics and stand-alone laser scanning.