Alarmingly high numbers of bats are being killed at wind turbines worldwide, raising concerns about the cumulative effects of bat mortality on bat populations. Mitigation measures to effectively reduce bat mortality at wind turbines while maximising energy production are of paramount importance. Operational mitigation (i.e. feathering wind turbine rotors at times of high collision risk for bats) is currently the only strategy that has been shown to substantially reduce bat mortality. This study presents a model based approach for developing curtailment algorithms that account for differences in bat activity over the year and night-time and are specific to the activity level at a certain wind turbine. The results show that easily measurable variables (wind speed, month, time of night) can predict times of higher bat activity with a high temporal resolution. A recently published collision model that was developed based on an excessive carcass search study is then applied to predict bat collision rate based on the modelled bat activity. Using the ratio of wind energy revenue and collision rate, 10 min intervals were weighted, so that turbines are stopped when collision rate is high and loss in revenue is low. A threshold of two dead bats per year and turbine resulted in a mean loss in annual revenue of 1.4%. The presented approach of acoustic monitoring at the nacelle and turbine specific curtailment has become the standard method to mitigate collision risk of bats at wind turbines in Germany.