This one-group pre-post test design was to evaluate sound distribution and sudden peak noise frequencies (SPNs) and the associated events after using a noise-sensor light alarm in a tertiary neonatal intensive care unit (NICU). The alarm is activated as the sound level reaches>or=65 A-weighted decibel (dBA). The environmental sound level was monitored continuously for a period of 1 week before and 1 month after using the alarm. The mean sound level in the incubator of patients receiving ventilator support before and after using the device were 58.0+/-0.6 and 56.4+/-0.7 dBA (t=8.619; p<0.001), whereas those at the radiant heated bed were 58.0+/-2.4 and 58.1+/-2.0 dBA (t=0.715; p=0.476). The percentage of observation time of sound levels<58 dBA increased by 28% in the incubator and 4% at the radiant heated bed (p<0.001). Episodes of SPN decreased from 630 to 185 times/d in the incubator and from 2069 to 748 times/d at the radiant heated bed after using the device. The noise-sensor light alarm effectively reduces sound level and episodes of SPN in the NICU. This may alleviate stress of noise for newborns with critical illness.