Investigating individual behavioural variations in mammals is essential for understanding their ecology and evolution, and plays a critical role in conservation and management practices. However, the reconstruction of long-term individual behaviour, such as via bio-logging, remains challenging owing to cost constraints and limitations of battery life and the impact of device size for smaller animals. This study proposed and tested a novel and cost-effective method for retrospective isotope analysis using mammalian eye lenses, specifically focusing on brown bears (Ursus arctos). Seven pairs of bear eye lenses were collected from southwestern Hokkaido, Japan. One or both lenses of each bear were segregated into small fragments from the outermost to the core tissues, and the nitrogen and carbon stable isotope ratios (δ15N and δ13C) of each eye lens fragment were measured. These isotope ratios in the eye lenses were compared with the δ15N and δ13C of potential brown bear diets in the study region (C3 herbs, C3 fruits, terrestrial animals, and corn). Additionally, we compared the isotopic patterns in both the right and left lenses of the same bear in two individuals to evaluate the consistency of our preparation protocol. In all eye lenses, high δ15N values were identified near the core, which gradually decreased towards the outer tissues, indicating ontogenetic dietary shifts related to lactation and weaning in the early life stage. Bears from study areas with high corn availability exhibited substantial increases in δ13C and δ15N in the outer lens tissues, suggesting a dietary shift toward corn consumption after weaning. Isotopic patterns between lens pairs from the same individual were similar, although discrepancies increased in tissues located 1.00 to 2.25 mm from the core, highlighting the need for standardisation in sample processing. This study demonstrates a novel and simple technique for retrospective isotope analysis in wild mammals using eye lenses, effectively reconstructing the feeding histories of brown bears. Our findings provide a new avenue for studying individual time-series behavioural patterns, with important implications for the fields of mammalian ecology, evolution, conservation, and management.