Lighting systems in commercial buildings are major contributors to the electricity consumption while occupants of these buildings are usually not aware of the effect of their energy-related behavior on the electricity consumption. A non-intrusive load disaggregation approach for lighting systems in office buildings was proposed and evaluated to provide high granularity spatiotemporal load monitoring. In our approach, we proposed a process for artificial light event detection and power consumption estimation using light intensity signals and rooms' contextual information (i.e. room area and number of possible states for lighting fixtures). Three event detection algorithms: generalized likelihood ratio test, signal-shape driven event detection, and enhanced naive event detection were proposed and evaluated. The signal-shape driven event detection algorithm outperformed on data collected in rooms where the natural light was used in conjunction with artificial light source. The correlation between the illuminance in a room and power consumption associated with that level of illuminance was used for power consumption estimation. The evaluations in rooms (with and without natural light) showed the feasibility of accurate power estimation when the sensor is installed closer to the floor and the estimation is informed by the possible operating states of lighting fixtures. (C) 2013 Elsevier B.V. All rights reserved.