Background: Many intensive care unit (ICU) patients describe noise as stressful and precluding sleep. No previous study in the adult setting has investigated whether room size impacts sound levels or the frequency of disruptive sounds. Methods: A-frequency S-time weighted equivalent continuous sound (L(AS)eq), A-frequency S-time weighted maximum sound level (L(AS)max) and decibel C peak sound pressure (L(C)peak) were measured during five 24-h periods in each of the following settings: three-bed room with nursing station (NS) alcove, single-bed room with NS alcove (1-BR with NSA) and single-bed room with bedside NS. Cumulative restorative time (CRT) (>5min with L(AS)max <55dB and L(C)peak <75dB) was calculated to describe calm periods. Two 8-h bedside observations were performed in each setting in order to note the frequency and sources of disruptive sounds. Results: Mean sound pressure levels (L(AS)eq) ranged between 52 and 58dBA, being lowest during night shifts. There were no statistically significant differences between the room types in mean sound levels or in CRT. However, disruptive sounds were 40% less frequent in the 1-BR with NSA than in the other settings. Sixty-four percent of disruptive sounds were caused by monitor alarms and conversations not related to patient care. Conclusions: Single-bed rooms do not guarantee lower sound levels per se but may imply less frequent disruptive sounds. Sixty-four percent of disruptive sounds were avoidable. Our findings warrant sound reducing strategies for ICU patients.