Though Typical Meteorological Year (TMY) data is widely used for predicting daylight quality and daylight-responsive control system performance, few studies investigated whether the predictions were accurate. In this study, daylight simulation was performed for a real space using the TMY data and an actual local weather data with a one-minute interval. The former was used to predict the daylight quality, to design a closed-loop daylight-responsive dimming control system, to select the calibration hours, and to predict the control system performance; the latter was used to characterize the actual daylight quality and control system performance. Though the actual weather conditions can provide 30% higher daylight levels to the space, which seemed to suggest a potential to achieve a greater energy saving, none of the existing CBDM measures suggested such a difference. The TMY data was used to design a closed-loop dimming control system, to select the appropriate weather conditions for system calibration, and to predict the energy saving of the system by limiting the occurrence of over-dimming below 2% of the occupied hours. It was found the predicted energy saving was generally accurate if the system calibration was performed when the weather conditions were similar to those identified using the TMY data. When the calibration was performed under the inappropriate weather conditions, the system achieved lower energy saving and had a much frequent occurrence of over-dimming conditions. In addition, performing the system calibration under the weather conditions that resulted in higher photosensor signal to work-plane illuminance ratios may be helpful to reduce the occurrence of over-dimming conditions caused by the larger variation of the weather conditions in reality. (C) 2018 Elsevier B.V. All rights reserved.