Participants (N=40) occupied a glare-free, daylit office laboratory for 1 day, and were prompted every 30 min to use dimming control over electric lighting to choose their preferred light level. Illuminances and luminances were recorded before and after each control opportunity; luminance maps were generated using a calibrated, high-dynamic range digital camera. Although there was a wide variation in chosen light levels between individuals, results showed a significant negative correlation between prevailing desktop illuminance and change in dimmer setting. This indicates that, from the perspective of occupants, daylight does displace electric lighting. Surprisingly, we did not find any luminance-based measure that was as good a predictor of participant dimmer choice as illuminance measured on the desktop. On average, manual dimming control in this situation reduced energy use for lighting by 25% compared to a fixed system delivering 500lx of electric lighting on the desktop.