This chapter was written during a moment of flux in the unravelling of the plaques, tangles, and tales that characterise over thirty years of research for treatments of dementia involving humans and animals. In the autumn of 2022, there were widely circulated reports in the mainstream media of a pharmaceutical breakthrough following long periods of investment in the so-called ‘amyloid cascade hypothesis’ (ACH). This focused on the small benefits in slowing cognitive decline for some people using the drug lecanemab. While modest, the results were reported as a significant success: ending ‘decades of failure’ and promising a step towards ‘a new era of drugs to treat Alzheimer’s’ (BBC News, 2022). However, even amid celebrations there were cautions about the wider potential of this drug, which could only be given to people with early Alzheimer’s disease who had evidence of amyloid plaques from scans or spinal fluid analysis, excluding the 98% of people who were untested, many other types of dementia, and those in the later stages of disease (Walsh et al., 2022). The success of lecanemab had been constructed, or fabricated, through careful articulation of the relations between one version of dementia, the associated animal models, targeted drug pathways, and a particular patient population. Yet, this highly specific ‘scientific breakthrough’ (Brown, 2000) achieved extensive celebratory media currency after many years of unfulfilled expectations for all those hoping for a cure for dementia. We use the story of this scientific breakthrough to frame our exploration of the organisation of translational research on dementia using mouse models of Alzheimer’s disease. We are interested in how the trajectories of drug development tightly knit together particular animals, humans, and drug pathways by focusing on the specific biological processes that can be modelled, evidenced, and altered in multi-species experimental systems (Davies, 2010, 2013b; Milne, 2020). We are also interested in the forms of dementia and other ways of doing research that fall outside of these biological pathways. While the therapeutic breakthroughs enabled by the ACH are to be celebrated, they raise a challenging issue for constructing multispecies dementia in biomedical research. Bringing animal models in the laboratory and humans in the clinic into sufficiently close relations so that manipulations in one organism effectively model processes relevant to therapeutic outcomes in another means bracketing off alternative research options and the multiplicity of human experiences of dementia. Making decisions about which animal model to progress has implications for both the animals used in research and the humans affected by this health condition. We are interested in how we might tell stories about research and drug development that are more accountable to the multiplicity of people and animals’ lives they touch. We use the concept of how mouse models are ‘fabricated’ in dementia research to focus attention on these processes of decision-making in science. We use this concept to unpack how disease models are fabricated as a way of bringing the decisions made by researchers in specific research trajectories into conversation with the decisions made by people affected by dementia who are reviewing new research proposals for medical charities. We explain our use of the concept of fabrication more fully below but note here that we consider fabrication an important process in all scientific research in which decisions to focus on certain things give shape to the tools, techniques, animals, and evidence that are built into research pathways. Yet we use the term acknowledging it can have a pejorative meaning too. These processes are fallible and open to other kinds of fabrication and forgery that reflect motivations and methods in science; though malfeasance is only a small part of the uncertainties around reproducible and responsible science (Macleod and the University of Edinburgh Research Strategy Group, 2022). The stories of dementia research contain irresponsible behaviour as well as successful breakthroughs for science; repeatedly deferred hope as well as expectations realised for people affected by dementia. We start by discussing our use of the concept of fabrication and introduce the research we draw on in this chapter. We then present two different ways of telling stories about the organisation of multispecies biomedical research on dementia. First, we trace the experimental practices used to construct mouse models within the ACH. This focuses on how researchers fabricate the animals and patient populations that link and make visible the specific biological pathways they hope will make a difference to disease trajectories. Second, we explore how people affected by dementia, acting as patient representatives in reviewing proposals for new biomedical research, locate themselves and their families within these stories of recurrent hope and delayed promise. We explore how they relate to the animals they encounter in research proposals and their responsibilities to others affected by dementia. While the future biomedical research trajectories of the ACH are underpinned by the translations made possible by transgenic animals based on specific family genomes, the patient representatives we spoke to wanted to acknowledge a wider set of relations, drawing attention to the importance of equity and balance to benefit those affected by Alzheimer’s disease now. We conclude by reflecting on the different dimensions of accountability in dementia research and the challenge of working between multispecies relations and the multiple versions of dementia.