In this study, four metaheuristic algorithms including particle swarm optimization (PSO), genetic algorithm (GA), differential evolution (DE) and simulated annealing (SA) were used for one-, two- and three-dimensional (1D, 2D and 3D) geophysical inverse problems. Theoretical and/or field data sets obtained by self-potential (SP), direct current resistivity (DCR), magnetic and crosshole radar applications were interpreted by one of the above-mentioned metaheuristics. PSO was used to determine model parameters (i.e., the electric dipole moment, polarization angle, depth, shape factor and origin of the anomaly) of SP anomalies which are both synthetically generated and measured over a graphite deposit in the southern Bavarian woods, Germany. A real-valued GA was used for estimating the parameters of a horizontally-layered earth model (i.e., resistivity and thickness of each layer) from vertical electrical sounding curves via the data sets based on both theoretical and a field experiment in a karstic environment in Bozdağ, İzmir (Turkey). A synthetic crosshole radar data set was considered for 2D imaging of the subsurface radar velocity distribution by a hybrid approach based on sequential use of SA and a linearized smoothness-constrained least-squares scheme, and DE algorithm was applied for a 3D inversion of a synthetically produced total field magnetic anomaly map. User-defined parameters required by each metaheuristic algorithm were determined by test studies considering the problems studied. Confidences in the results obtained by the metaheuristics were also examined by various uncertainty and statistical analyses. Since the metaheuristics used here produced satisfactory results for estimating the model parameters of a variety of the geophysical problems, it can be concluded that these algorithms can be applied to low- and relatively high-dimensional geophysical data.