Societal concerns over the potential impacts of recent global change have prompted renewed interest in the long-term ecological monitoring of large ecosystems. The deep sea is the largest ecosystem on the planet, the least accessible, and perhaps the least understood. Nevertheless, deep-sea data collected over the last few decades are now being synthesised with a view to both measuring global change and predicting the future impacts of further rises in atmospheric carbon dioxide concentrations. For many years, it was assumed by many that the deep sea is a stable habitat, buffered from short-term changes in the atmosphere or upper ocean. However, recent studies suggest that deep-seafloor ecosystems may respond relatively quickly to seasonal, inter-annual and decadal-scale shifts in upper-ocean variables. In this review, we assess the evidence for these long-term (i.e. inter-annual to decadal-scale) changes both in biologically driven, sedimented, deep-sea ecosystems (e.g. abyssal plains) and in chemosynthetic ecosystems that are partially geologically driven, such as hydrothermal vents and cold seeps. We have identified 11 deep-sea sedimented ecosystems for which published analyses of long-term biological data exist. At three of these, we have found evidence for a progressive trend that could be potentially linked to recent climate change, although the evidence is not conclusive. At the other sites, we have concluded that the changes were either not significant, or were stochastically variable without being clearly linked to climate change or climate variability indices. For chemosynthetic ecosystems, we have identified 14 sites for which there are some published long-term data. Data for temporal changes at chemosynthetic ecosystems are scarce, with few sites being subjected to repeated visits. However, the limited evidence from hydrothermal vents suggests that at fast-spreading centres such as the East Pacific Rise, vent communities are impacted on decadal scales by stochastic events such as volcanic eruptions, with associated fauna showing complex patterns of community succession. For the slow-spreading centres such as the Mid-Atlantic Ridge, vent sites appear to be stable over the time periods measured, with no discernable long-term trend. At cold seeps, inferences based on spatial studies in the Gulf of Mexico, and data on organism longevity, suggest that these sites are stable over many hundreds of years. However, at the Haakon Mosby mud volcano, a large, well-studied seep in the Barents Sea, periodic mud slides associated with gas and fluid venting may disrupt benthic communities, leading to successional sequences over time. For chemosynthetic ecosystems of biogenic origin (e.g. whale-falls), it is likely that the longevity of the habitat depends mainly on the size of the carcass and the ecological setting, with large remains persisting as a distinct seafloor habitat for up to 100 years. Studies of shallow-water analogs of deep-sea ecosystems such as marine caves may also yield insights into temporal processes. Although it is obvious from the geological record that past climate change has impacted deep-sea faunas, the evidence that recent climate change or climate variability has altered deep-sea benthic communities is extremely limited. This mainly reflects the lack of remote sensing of this vast seafloor habitat. Current and future advances in deep-ocean benthic science involve new remote observing technologies that combine a high temporal resolution (e.g. cabled observatories) with spatial capabilities (e.g. autonomous vehicles undertaking image surveys of the seabed).