The lack of rotational inertia requires fast action for frequency control, thus measurements delays can create instabilities or make the controller ineffective. This is a challenge for grid-connected Microgrids whose frequency requires to be adequately monitored and maintained under operationally margins. Generally, in a power system, the strongest generators are the ones capable of maintaining the frequency of the system, however, when having distributed resources it is harder to keep such frequency between a secure threshold. Therefore, estimating frequency values in a short period of time becomes a helpful feature in this new structure of power systems. This work discusses the Extended Kalman Filter (EKF) application for grid frequency estimation in AC connected microgrids by using a simple second order model. Furthermore, the EKF is used for forecasting the nadir according to the sample time variation. A microgrid benchmark is developed and tested under several load scenarios to test the frequency estimation process.