Time-dependent variation and mechanism of asphalt-aggregate interface features under ultraviolet (UV) aging have not yet been illustrated. This study explored the time-dependent variation and mechanism of asphalt-aggregate interface features under UV aging. To this end, six asphalt binders were chosen and subjected to different durations of UV aging, and then the variation behaviors of their cohesion-adhesion ability with the changing UV radiation time were analyzed through surface energy theory. Further, the microscopic attributes of asphalt binders experiencing the various UV aging durations were measured and their varying rules were investigated with a series of micro observations. Finally, the relationship between cohesion-adhesion characteristics and microscopic properties were illustrated and thus the variation mechanism of interfacial cohesion and adhesion were revealed by mathematical statistics and image presentation. The results showed that UV aging weakened the cohesion and multi-working adhesion of asphalt binders and the moisture stability of asphalt-aggregate systems, and overall presented a gradually strengthened influence on them with the prolonged UV radiation time, especially the UV radiation exceeded 6 days. The variation of phase morphology of asphalt binders under UV aging depended on the asphalt types and neat asphalt oil sources, and there existed a favorable linear relationship between Catana phase of asphalt binders and surface roughness. Adhesion force, function group and four components (saturate (S), aromatics (Ar), resin (R) and asphaltene (As)) also exhibited a gradually increasing variation as the UV radiation duration went on, and among which, the UV aging reduced the adhesion force of asphalt binders. Adhesion force Fad demonstrated the highest Pearson relevance with the cohesion and aqueous-condition adhesion and the moisture stability of asphalt-aggregate system, and the cohesion-adhesion properties of asphalt binders under UV aging met the multivariate linear mathematical relationship with the microscopic attributes screened, with the goodness-of-fit greater than 0.7000.