Appropriate end-points are integral to the refinement of laboratory animal experiments. Our recent experience has highlighted that ambiguity around end-points is hampering their adoption in experiments that cause severe suffering to fish. In toxicology, the term endpoint (single word) refers to the response variable to the treatment that is measured and analysed. This differs to usage within laboratory animal experimentation, where end-point (hyphenated) refers to the time-point when exposure of the animal(s) to the treatment (and suffering) ends. Within laboratory animal experimentation, standardised terminology is needed for different types of early end-point which are based on the condition of the animal(s) or progress of the experiment. We propose that those involved in regulating and conducting animal experiments consider seven distinct types of early end-point (aim, technical error, biological error, mortality, moribundity, prognostic humane, non-prognostic humane) in addition to the planned experimental end-point (i.e. maximum duration). Moribundity (not morbidity) refers to an animal in a severely debilitated state close to death. Moribundity in fish is not yet defined, so we propose identification via a lack of response to external stimuli, loss of equilibrium (i.e. loss of righting reflex), and a slow opercular ventilation rate. As these clinical signs equate to those of deep/surgical anaesthesia, this moribundity end-point cannot be considered a humane end-point as the fish is likely to be unconscious and have passed the point of maximum suffering. We believe that identification of earlier humane end-points based on clinical signs and wider recognition of other types of early end-point can reduce suffering in experiments.