The invention provides an LIBS (Laser-induced Breakdown Spectroscopy) quantitative gold ore method based on an ultraviolet band and a model stacking strategy, which belongs to the technical field of data mining and comprises the following steps: S1, acquiring spectral data by using an LIBS spectrometer with the ultraviolet band, and establishing a data set; s3, spectral data preprocessing and spectral feature extraction; s4, selecting and establishing a base learner model and a meta learner model; s5, establishing a quantitative inversion model by adopting a model stacking strategy; and S6, result verification. The method mainly comprises the following steps: carrying out quantitative analysis research on gold elements on natural gold ore rocks collected in a gold ore exploration process, carrying out crushing, ball-milling and tabletting treatment on the samples in a laboratory, collecting spectral data by adopting an LIBS (Laser-induced Breakdown Spectroscopy) instrument with an ultraviolet band, and establishing a data set; in order to scientifically and accurately quantify the gold element, inhibit the matrix effect and the self-absorption effect to the greatest extent and output and verify a quantitative prediction result, the final result is high in prediction precision, strong in anti-interference capability and good in model generalization.