At the design stage of bridges, all possible actions and their combinations are to be considered. In certain cases, the influence of the environment must be taken into account in addition to design values of traffic loads. In order to assess the current state of an existing bridge, actual applied actions must be considered: updated traffic situation, monitored climatic actions, and their unfavorable combinations. Therefore, monitoring all actions makes it possible to adequately study a structure. Since only limited data are generally available, the important question is how the quality and the duration of monitoring influence the assessment of the structure. In the current study applied to the Millau viaduct, effects from monitored traffic and wind actions are evaluated. The statistical analysis of applied actions and caused effects is done according to the Peaks Over Threshold (POT) approach. Results include the comparison between confidence intervals of predictions, for each studied load case and for various periods of monitoring. In addition, this paper presents study of the influence of the length of monitoring data on predictions of future extreme load cases, and also proposes an alternative efficient algorithm for threshold choice in the POT approach.