This study validated an UPLC-MS/MS analysis method to simultaneously detect twelve cannabinoids. The concentration-response relationship for all analyzed cannabinoids was linear with R ² values > 0.99 using the developed method, the relative standard deviations of recovery for the three levels of spiked samples are between 66.1%~104.1%. The method was used to analyze 43 industrial hemp flower and leaf samples, with the data being statistically analyzed. The analysis results showed that the cannabinoids content varied significantly among different varieties and different pretreatment methods, sample heating treatment led to the decarboxylation of acidic cannabinoids and the mutual conversion of cannabinoids. Based on the statistical analysis of the cannabinoids, hemp from different regions and different varieties were well distinguished by the PLS-DA model, with the main contributing substances being Cannabidiol, Δ ⁹-tetrahydrocannabinol, and Δ ⁸-tetrahydrocannabinol.