Skin sensitization involves a two-phase process where exposure to haptens triggers an immune response. Predicting skin sensitization is crucial for identifying potential hazards in substances applied topically or ingested. Traditional methods, such as the guinea pig maximization test and local lymph node assay (LLNA), have relied on animal testing, raising ethical concerns and regulatory pressure for alternative methods. These traditional tests are invasive, time-consuming and sometimes not directly relevant to human skin responses. Recent advancements have introduced several non-animal models. In vitro methods use cell cultures or reconstructed tissues to simulate immune responses. Ex vivo methods utilize isolated animal tissues, while in chemico methods focus on chemical reactivity assays like the direct peptide reactivity assay (DPRA). In silico models leverage computational approaches, including the Derek Nexus, OECD QSAR Toolbox, TOPKAT, Tissue Metabolism Simulator for Skin Sensitization (TIMES-SS) and Tox21. These models predict sensitization by analysing chemical structures and biological activities, reducing the need for animal testing. Integrated models, combining in vivo, in silico and in chemico approaches, offer a comprehensive strategy for toxicity prediction. Artificial intelligence (AI) and deep learning models like DeepTox further enhance predictive accuracy and efficiency, processing large datasets to identify potential sensitizers early in the development process.