Considering that natural products as tyrosinase inhibitors are considered to be safe, with little or no toxic side effects and friendly to the environment, it is urgent to develop a new recognition strategy for natural tyrosinase inhibitors. In current study, an integrated computational analysis was conducted on Cys-containing dipeptides with high tyrosinase inhibitory activity. Firstly, molecular fingerprint similarity (FS) clustering analysis was performed on the target molecule using machine learning. Secondly, genetic algorithm was used to construct two kinds of highly accurate QSAR models (R2 = .978 and .984, respectively) with Cys at C-terminal and N-terminal. Finally, three novel natural candidate inhibitors (NP1, NP2, NP3) were discovered using Molnatsim natural product cluster library, automated screening process and QSAR based on the maximum common substructure (MCS) algorithm, their IC50pre were 260.96, 3.37 and 0.05 μm/mol. Pharmacokinetic predictions showed that NP2 and NP3 had high Bioavailability Score (BS) and Gastrointestinal (GI) absorption, and molecular dynamics simulations further validated the stability of these novel natural candidate inhibitors in binding to tyrosinase. In conclusion, our results provide new ideas for discovering new activities of natural products, and provide an accurate QSAR model for developing novel tyrosinase inhibitors based on MCS Cys-containing dipeptides. PRACTICAL APPLICATIONS: Tyrosinase is related to the occurrence of diseases such as excessive melanin deposition such as freckles and chloasma, and studies have shown that neurodegeneration associated with Parkinson's disease and Huntington's disease is also related. In addition, enzymatic browning on the surface of fresh fruit and vegetable slices will shorten the shelf life and affect their quality. Therefore, screening, designing and developing efficient tyrosinase inhibitors is very important in the fields of medicine, cosmetics, food and so on.