The control of temperature during the acute phase of stroke may be a new therapeutic target that can be applied in all stroke patients, however therapeutic window or timecourse of the temperature effect is not well established. Our aim is to study the association between changes in body temperature in the first 72 hours and outcome in patients with ischemic (IS) and hemorrhagic (ICH) stroke. We prospectively studied 2931 consecutive patients (2468 with IS and 463 with ICH). Temperature was obtained at admission, and at 24, 48 and 72 hours after admission. Temperature was categorized as low (<36°C), normal (36–37°C) and high (>37°C). As the main variable, we studied functional outcome at 3 months determined by modified Rankin Scale.Temperature in stroke patients is higher than in controls, and increases gradually in the first 72 hours after stroke. A positive correlation between temperature and stroke severity determined by NIHSS was found at 24 and 48 hours, but not at admission or 72 hours. In a logistic regression model, high temperature was associated with poor outcome at 24 hours (OR 2.05, 95% CI 1.59–2.64, p<0.0001) and 48 hours (OR 1.93, 95% CI 1.08–2.34, p = 0.007), but not at admission or 72 hours.Temperature increases in patients with stroke in the first 72 hours, with the harmful effect of high temperature occurring in the first 48 hours. The neuroprotective effect of low temperature occurs within the first 24 hours from stroke onset.