Unlike wards, where chronic and acute pain are regularly managed, comparisons of the most commonly used self-report pain tools have not been reported for the intensive care unit (ICU) setting. The objective of this study was to compare the feasibility, validity and performance of the Visual Analog Scale (horizontal (VAS-H) and vertical (VAS-V) line orientation), the Verbal Descriptor Scale (VDS), the 0-10 oral Numeric Rating Scale (NRS-O) and the 0-10 visually enlarged laminated NRS (NRS-V) for pain assessment in critically ill patients. One hundred and eleven consecutive patients admitted into a medical-surgical ICU were included as soon as they became alert and were able to follow simple commands. Pain was measured using the 5 scales in a randomized order upon enrollment-(T1) and after-(T2) administration of an analgesic or, in absence of pain upon enrollment, after a nociceptive procedure. The rate of any response obtained both at T1 and T2 (success rate) was significantly higher for NRS-V (91%) compared with NRS-O (83%), VDS (78%), VAS-H (68%) and VAS-V (66%). Pain intensity changed significantly between T1 and T2, showing a good validity and responsiveness for the 5 scales, which correlated well between each other. The negative predictive value calculated from true and false negatives defined by real and false absence of pain was highest for NRS-V (90%). In conclusion, the NRS-V should be the tool of choice for the ICU setting, because it is the most feasible and discriminative self-report scale for measuring critically ill patients' pain intensity.