Gilthead seabream were equipped with intraperitoneal biologging tags to investigate cardiac responses to hypoxia and warming, comparing when fish were either swimming freely in a tank with conspecifics or confined to individual respirometers. After tag implantation under anaesthesia, heart rate (f(H)) required 60 h to recover to a stable value in a holding tank. Subsequently, when undisturbed under control conditions (normoxia, 21 degrees C), mean f(H) was always significantly lower in the tank than in the respirometers. In progressive hypoxia (100% to 15% oxygen saturation), mean f(H) in the tank was significantly lower than in the respirometers at oxygen levels down to 40%, with significant bradycardia in both holding conditions below this level. Simultaneous logging of tri-axial body acceleration revealed that spontaneous activity, inferred as the variance of external acceleration (VAR(m)), was low and invariant in hypoxia. Warming (21 to 31 degrees C) caused progressive tachycardia with no differences in f(H) between holding conditions. Mean VAR(m) was, however, significantly higher in the tank during warming, with a positive relationship between VAR(m) and f(H) across all temperatures. Therefore, spontaneous activity contributed to raising f(H) of fish in the tank during warming. Mean f(H) in respirometers had a highly significant linear relationship with mean rates of oxygen uptake, considering data from hypoxia and warming together. The high f(H) of confined seabream indicates that respirometry techniques may bias estimates of metabolic traits in some fishes, and that biologging on free-swimming fish will provide more reliable insight into cardiac and behavioural responses to environmental stressors by fish in their natural environment.