Reverse alarms should be clearly audible and elicit sufficient reaction, without being excessively loud. Unfortunately, no one set of guidelines is consistently used in the field to ensure adequate alarm levels. Detection thresholds and reaction thresholds were measured for two reverse alarms (tonal and broadband) in young adults with normal hearing in 80-dBA workplaces noises, with and without the use of passive and level-dependent hearing protection devices (HPDs). Reverse alarms remained audible at levels well below the background noises, with and without the selected HPDs (thresholds ranging from -13 to -24 dB S/N across both alarms), while a reaction was elicited at levels up to 7 dB above the background noise in some conditions of HPD use. Compared to the tonal alarm, the broadband alarm generally yielded higher detection thresholds but lower reaction thresholds. Detection was more influenced by alarm and noise type, while reaction was more sensitive to HPD use. Indeed, reaction thresholds were always higher with hearing protection compared to without, although level-dependent protection yielded reaction thresholds closer to unprotected values than passive protection in some cases. Since noise type appears to have a more limited effect on reaction thresholds, adjusting alarms based on a global dBA method such as ISO 9533 (SNR ≥ 0 dB) appears suitable. However, while the alarm adjustment criterion recommended in ISO 9533 seems adequate for unprotected listening for individuals with normal hearing, an additional amount (up to 7 dB) may be warranted to elicit the same reaction when HPDs are used, particularly for passive hearing protection. In addition, the definition of background noise in ISO 9533 should be extended to include all noise sources in the vicinity of the reversing vehicle.