Global warming is driving climate change to levels not experienced since the advent of agriculture, primarily due to anthropogenic factors and the accumulation of CO2. Rising temperatures, frequent droughts, and elevated CO2 levels are reducing crop productivity in key agricultural regions. Developing climate-resilient crop varieties is essential. Metabolomics provides a powerful tool for quantifying plant responses to abiotic stressors and identifying predictive biomarkers for stress tolerance. While metabolite-based diagnostics are well-established in clinical research, their integration into crop breeding remains limited. This Tansley review highlights recent advances in metabolomics for predicting yield stability and quality under stress, emphasising the role of metabolic biomarkers in resolving complex genotype 9 environment interactions. We discuss the utility of metabolite quantitative trait loci, metabolome-wide association studies, and machine learning-driven metabolic marker-assisted genomic prediction in enhancing trait prediction. These approaches complement genomic selection, improving accuracy and resilience forecasting. We also address methodological challenges in translating metabolomics into breeding pipelines, including standardisation and data integration. By combining metabolomics with genomics, modelling, and high-throughput phenotyping, researchers can accelerate the development of stress-resilient crops. This Tansley review presents a framework for leveraging metabolomics in predictive breeding, offering a transformative pathway toward sustainable agriculture in a changing climate.
