Changes in the size and depth of sources greatly affect self-potential (SP) anomalies. Therefore, it is important to determine the location of the source accurately. In the present study, applications of the normalized full gradient (NFG) method and Euler deconvolution (EUD) were described to determine the location of the sphere-like SP body as complementary approaches to other optimization algorithms. The NFG and EUD methods were tested on synthetic, noise-free, and noisy anomalies caused by sphere-like models in two-dimensional (2D) and three-dimensional (3D) cases. Subsequently, the methods were applied to real field data. The importance of the present study lies in the fact that it is the first 3D application of these methods to the SP anomaly caused by the sphere-like model in the literature. In order to determine the optimum harmonic number in the NFG method, a new criterion was used instead of the usual trial-and-error method, providing more reliable selection possibilities. In a similar way, average values were used to determine the window size accurately in the EUD method. The test results of the synthetic and real field models were satisfactory. They showed that both methods are applicable to determine the location of sphere-like structures, such as ore deposits, in self-potential surveys.