For the purpose of estimating the state of a linear time-invariant system with measurements subject to outliers, we propose an observer with a saturated output injection in such a way to mitigate the effect of abnormal and isolated measurement noise on the error dynamics. Stability conditions in both the continuous-time and the discrete-time cases are derived, which ensure global exponential stability to the origin for the error dynamics. Such conditions can be expressed in terms of linear matrix inequalities, allowing for a viable design by using convex optimization. The effectiveness of the approach is illustrated by means of simulations in comparison with the Luenberger observer.