Studies with monochromatic light stimuli have shown that the action spectrum for melatonin suppression exhibits its highest sensitivity at short wavelengths, around 460 to 480 nm. Other studies have demonstrated that filtering out the short wavelengths from white light reduces melatonin suppression. However, this filtering of short wavelengths was generally confounded with reduced light intensity and/or changes in color temperature. Moreover, it changed the appearance from white light to yellow/orange, rendering it unusable for many practical applications. Here, we show that selectively tuning a polychromatic white light spectrum, compensating for the reduction in spectral power between 450 and 500 nm by enhancing power at even shorter wavelengths, can produce greatly different effects on melatonin production, without changes in illuminance or color temperature. On different evenings, 15 participants were exposed to 3 h of white light with either low or high power between 450 and 500 nm, and the effects on salivary melatonin levels and alertness were compared with those during a dim light baseline. Exposure to the spectrum with low power between 450 and 500 nm, but high power at even shorter wavelengths, did not suppress melatonin compared with dim light, despite a large difference in illuminance (175 vs. <5 lux). In contrast, exposure to the spectrum with high power between 450 and 500 nm (also 175 lux) resulted in almost 50% melatonin suppression. For alertness, no significant differences between the 3 conditions were observed. These results open up new opportunities for lighting applications that allow for the use of electrical lighting without disturbance of melatonin production.