A different approach for attitude filtering integrates the filter with one of the static attitude estimators (e.g. QUEST) such that the filter uses the estimated attitude by this estimator as the measurement. Among several advantages such as being robust and providing fast convergence, this architecture is also useful when several measurements are integrated. In contrast, this approach cannot provide any measurement to the filter when only one vector measurement is available. In this study, a cascaded static attitude estimator and Kalman filter architecture is proposed and it is shown that coarse attitude estimates are possible using only the magnetometer measurements. Results show that the proposed method can estimate the attitude with an accuracy better than 5 deg, specifically if the satellite is not experiencing high angular rates and the sampling frequency for magnetometer measurements is high. Besides providing a standalone solution, the given method can be used as a back-up when a magnetometer-only attitude solution is needed, e.g. when the satellite is in eclipse.