The method of the present invention can be used to monitor and predict the condition of an automatic door of an elevator or more generally an automatic door in a building. In the method, the acceleration or velocity of the door is measured and a dynamic model of the door is created. Using the model, estimated values of acceleration or velocity of the door can be calculated as a function of unknown parameters. One of the unknown parameters is the frictional force acting on the door during movement. By utilizing the estimated acceleration or velocity as well as measured acceleration or velocity values, an error function is obtained, whose minimum value is found using an optimizer. The unknown parameters corresponding to the minimum value indicate the current condition of the door. On the basis of earlier measurement results, it is additionally possible to predict a point of time when a failure is likely to occur in the operation of the door. In addition to unknown force parameters, it is possible, using a genetic algorithm, to determine the operational condition of a door closing device as well.