Autobiographical memory (AM) for personally experienced events is characterised by vivid recollections of past happenings and contexts. Evidence from lesion and neuroimaging studies has demonstrated that AM retrieval is mediated by a network of structures, including medial prefrontal cortex, hippocampus, precuneus, retrosplenial and posterior parietal cortex. The hippocampus plays a crucial role in this network; hippocampal damage is associated with AM deficits, and this structure is preferentially engaged by AM retrieval. The characteristics of AM that determine its involvement have yet to be elucidated. The first three studies used functional neuroimaging to examine how different qualities of AMs modulate hippocampal engagement in healthy individuals. Study 1 demonstrated that recollective qualities of AMs (e.g., detail, emotionality and personal significance) are important predictors of hippocampal activation, independent of factors such as recency. Study 2 showed that temporal specificity is not a key determinant of hippocampal engagement as similar activation was seen when retrieving unique or repeated events, although other AM network nodes exhibited differential responses. Study 3 found that retrieval of different recollective qualities was associated with distinct sub-networks, centred on the hippocampus. Investigations of the neural correlates of AM impairments in patients with unilateral hippocampal damage were also conducted. Study 4 confirmed that both left and right temporal lobe epilepsy patients exhibit AM deficits, and the severity of impairment correlated with the degree of hippocampal atrophy. Study 5 demonstrated significant reductions in activity across the AM network, including the hippocampus. However, more severe hippocampal damage was associated with increased activation of residual hippocampal tissue, particularly contralateral to the seizure focus. Together, the findings presented suggest the hippocampus is a key structure supporting autobiographical recollection, likely integrating different recollective aspects of the memory. Further, the hippocampus appears to be the "hub" of the AM network, and when damaged, the network fails to engage normally, leading to diminished autobiographical recollection. Finally, this thesis demonstrates how the AM paradigm can be used with different populations and various neuroimaging analysis techniques (e.g., parametric modulation, functional and effective connectivity, linear structural measurements) to explore distinct questions about the nature of AM and its neural substrates.