Abstract: A method for predicting speech quality in a voice call between a mobile terminal and a mobile coupling node is disclosed. The method includes obtaining M values of a power parameter measured M times at different points in time over a first time interval of the duration of the call, where the power parameter is a parameter indicative of transmitted power in one of a first direction and a second direction of the call, the first direction being an uplink direction from the mobile terminal to the mobile coupling node, the second direction being a downlink direction from the mobile coupling node to the mobile terminal. The method also includes calculating a power factor for the call based on the M values of the power parameter, and, based on the calculated power factor, calculating a speech quality indicator for the other one of the first direction and the second direction.