Background. Pathfinding plays a big role in both digital games and robotics, and is used in many different ways. One of them is multi-goal pathfinding (MGPF) which is used to calculate paths from a start position to a destination with the condition that the resulting path goes though a series of goals on the way to the destination. For the most part research on this topic is sparse, and when the complexity is increased through obstacles that are introduced to the scenario, there are only a few articles in the field that relate to the problem.Objectives. The objective in this thesis is to conduct an experiment to compare four algorithms for solving the MGPF problem on six different maps with obstacles, and then analyze and draw conclusions on which of the algorithms is best suited to use for the MGPF problem. The first is the traditional Nearest Neighbor algorithm, the second is a variation on the Greedy Search algorithm, and the third and fourth are variations on the Nearest Neighbor algorithm. Methods. To reach the Objectives all the four algorithms are tested fifty times on six different maps of varying sizes and obstacle layout. Results. The data from the experiment is compiled in graphs for all the different maps, with the time to calculate a path and the path lengths as the metrics. The averages of all the metrics are put in tables to visualize the difference between the results for the four algorithms.Conclusions. The conclusions were that the dynamic version of the Nearest Neighbor algorithm has the best result if both the metrics are taken into account. Otherwise the common Nearest Neighbor algorithm gives the best results in respect to the time taken to calculate the paths and the Greedy Search algorithm creates the shortest paths of all the tested algorithms.