Aim To enhance current attempts to understand biodiversity patterns by using an historical ecology approach to highlight the over-riding influence of land-use history in creating past, current and future patterns of biodiversity in fragmented agricultural landscapes. Methods We develop an integrative conceptual framework for understanding spatial and temporal variations in landscape patterns in fragmented agricultural landscapes by presenting five postulates (hypotheses) which highlight the important role of historical, anthropogenic disturbance regimes. We then illustrate each of these postulates with examples drawn from fragmented woodlands in agricultural areas of south-eastern Australia, and discuss these findings in an international context. Location examples are drawn from agricultural areas in south-eastern Australia. Results We conclude that there is limited potential to refine our understanding of patterns of biodiversity in human-modified landscapes based on traditional concepts of island biogeography, or simple assumptions of ongoing destruction and degradation. Instead, we propose that in agricultural landscapes that were largely cleared over a century ago: (1) present-day remnant vegetation patterns are not accidental, but are logically arrayed due to historic land-use decisions, (2) historic anthropogenic disturbances have a major influence on current ecosystem conditions and diversity patterns, and (3) the condition of remnant ecosystems is not necessarily deteriorating rapidly. Main conclusions An historical ecology approach can enhance our understanding of why different species and ecosystem states occur where they do, and can explain internal variations in ecological conditions within remnant ecosystems, too often casually attributed to the 'mess of history'. This framework emphasizes temporal changes (both past and future) in biotic patterns and processes in fragmented agricultural landscapes. Integration of spatially and temporallyexplicit historical land-use information into ecological studies can prove extremely useful to test hypotheses of the effects of changes in landscape processes, and to enhance future research, restoration and conservation management activities.