Atlantis II is the only hydrothermally active of the five Central Red Sea deeps, hot new brine being supplied by a geyser spring. It is filled with two stratified anoxic brine layers, namely, the Lower Conductive Layer (LCL) at the bottom and the Upper Conductive Layer (UCL) above it. The other deeps were filled by brine overspill. Hydrological and geochemical data on the Atlantis II Deep show that, between 1965 and 1977, only negligible amounts of lower brine have spilled over towards Chain A Deep and that, in contrast, the lower brine component of the UCL has increased significantly. The heat-mass balance of the Atlantis II Deep lower brine between 1966 to 1977 is considered. Mass balance calculations are based on the published bathymetry of the Deep and on the measured rising rate of the interface between the UCL and the LCL, including a possible loss of mass out of the system. Heat balance equations involve (1) the conductive flow of heat at the bottom of the Deep, (2) the heat lost by conduction through the lower interface, (3) the flow of heat due to hot brine advection and (4) the heat lost by lower brine advection out of the LCL. The equation of Pitzer et al. (1984) [57], extrapolated above 300°C, is first used to calculate the C p 's of the hydrothermal brine: it predicts moderately increasing C p 's with temperature. Heat-mass balance equations, solved with this C p model, yield a spring temperature of 461°C for the time interval 1966–1977. This temperature is unrealistic as it is ≈ 70°C higher than 390°C, the boiling temperature of the lower brine at 220 bar. Besides, it is shown that none of the parameters in the calculation, other than C p can attain values as to decrease T s below 390°C. The modified equation of Born predicts a near exponential increase of the hydrothermal brine C p 's above 300°C. Spring temperatures of 342°C or 353°C are calculated with this C p model, depending whether the mass lost towards the UCL is taken into account or not. These temperatures compare favorably with the mean temperature of 330°C obtained for the trapping of fluid inclusions in epigenetic anhydrites from the discharge zone sediments [21]. Hence it is concluded that (1) the changes in the mass and temperature of the Atlantis II Deep lower brine between 1965 and 1977 cannot be interpreted in terms of a realistic spring temperature, unless the hydrothermal fluid has presented very high heat transport properties above 300°C; (2) this can be either in the stable region below the boiling point, as predicted by the equation of Born, or in the metastable region above 390°C: superheated liquids indeed present heat capacities which increase exponentially with temperature. Two periods in the recent hydrothermal activity of the Atlantis II Deep are distinguished: (1) From 1965 to 1976, the Deep has received an excess heat supply, e.g., compatible with the influx of superheated liquids. The spring flow rate, including the mass of lower brine expelled towards the UCL [29], was around 367 kg s −1 . (2) The hydrothermal activity then decreased from 1976 to 1979. The influx of fluids with abnormal heat transport properties was greatly reduced and the mean spring flow rate probably dropped to ≈ 280kg s −1 .