Two independent suites of hydrofracturing stress measurements were conducted in the top 600 m of the Iceland Research Drilling Project deep hole at Reydarfjordur, east Iceland. As indicated by the continuously extracted drill core, the tested section consists of tertiary subaerial tholeiitic lava flows cut by many basaltic dikes. The density of the basalt was used to estimate the vertical stress &sgr;V (in magapascals) at depth D (in meters): &sgr;V=0.027D. In the 33- to 270-m depth range the horizontal principal stresses (&sgr;Hmax and &sgr;Hmax) as calculated from hydrofracturing results increase gradually with depth but as a lower rate than the vertical stress: &sgr;Hmin=2.1+0.016D; &sgr;Hmax=3.3 +0.016D. In the 300- to 550-m depth range a major perturbation occurs with both horizontal stress magnitudes rapidly rising between 300 and 400 m and then uniformly decreasing between 400 and 550 m to a level apparently predictable by the relationships obtained in the top 270 m. Conditions favoring normal faulting (&sgr;V &sgr;Hmax &sgr;Hmax) were established in the 540- to 580-m range and are conjectured to persist at greater depths since the vertical stress gradient is higher than that of &sgr;Hmax and &sgr;Hmin. This would be in accord with the predominant fault type in east Iceland. The relative low horizontal stress values and the small difference between them are supportive of ocean floor spreading theories and contradict near-surface high compressive stresses previously determined by Hast (1973) in east Iceland. The large increase in horizontal stress in the range 300- to 500-m depth is not understood. A correlation appears to exist between the zone of almost unintruded thick lava flows (200--500 m) and that of high stresses. An intuitive model suggesting 'locked in' stresses in this zone is proposed. The hydrofracturing-determined overall mean direction of &sgr;Hmax is N50¿E, while a more selective sampling of results yields N40¿E. The measured stress directions appear to be reasonably compatible with the local structure such as dike direction (NNE) and lava dip (WSW). Significantly, &sgr;Hmax direction is subparallel to the N20¿E trend of the nearest fissure swarms in the axial rift zone. This almost coincidence of directions, like the measured stress magnitudes, is also in agreement with the model of spreading lithospheric plates being pulled apart.