Hennessy and colleagues experimentally separate nostalgia from familiarity, a distinction often blurred in research on personally meaningful music. Fifty-seven younger and older adults listened during fMRI to three individually constructed categories: personalised nostalgic music, familiar non-nostalgic music and unfamiliar non-nostalgic music, with stimuli matched for musical features through a machine-learning-based procedure. Nostalgic music produced stronger activity than both control conditions across default-mode, salience, reward, medial-temporal and supplementary-motor regions, together with increased connectivity between self-referential and affect-related areas. Older adults showed stronger BOLD responses in nostalgia-related regions, while the predictors of neural response differed by age. The study’s key conceptual operation is to demonstrate that nostalgia cannot be reduced to recognition or familiarity; it constitutes a self-relevant affective-memory configuration. By combining personalised stimulus construction with acoustic matching, the paper bridges autobiographical memory, affective neuroscience and lifespan research, offering a rigorous model for isolating biographical charge from mere musical familiarity.