Determining the position of interaction is of great interest for gamma-ray imaging in various nuclear applications. Among all gamma-ray detectors, scintillation detectors are commonly exploited for imaging purposes because they can be prepared in large dimensions and are economically affordable. In this work, the general shape of the measured gamma-ray spectra of two long and large-area plastic scintillation detectors are analyzed by artificial neural networks to determine the position of interaction in one and two dimensions (1D and 2D), respectively. The position of interaction was treated as the position of a 137Cs gamma-ray point source on the long and large-area scintillation detectors. Utilizing this method, only one photomultiplier tube (PMT) was used for 1D positioning of interaction in a 4 × 4 × 35-cm3 long plastic detector, while just two PMTs were applied for 2D positioning of interaction in a 50 × 50 × 5-cm3 large-area plastic detector. The position of interaction in the long detector was determined with a resolution of 1 cm and a mean absolute error of less than 1%, while a resolution of 5 cm with a mean absolute error of 13% was achieved for the large-area detector.