There is controversy over the incidence of awareness during anesthesia. A review of 28 recent studies suggests that the incidence in 297,957 participants is 1:863 (0.12%),1 a number consistent with a large scale prospective study showing an incidence of 1:770 (0.13%).2 A much lower incidence of 1:14,560 (0.0068%) was reported in a recent retrospective analysis of a large continuous quality improvement (CQI) database.3 In this issue of the journal, Mashour et al.4 also describe a low incidence of self-reported awareness in their CQI database of 1:4401 (0.023%). Remarkably, this incidence was not significantly different from the incidence of undesired intraoperative awareness in those patients who did not receive general anesthesia. They conclude that retrospective analysis of a CQI database has inadequate resolution to study the incidence of awareness, a view with which I concur. Most prospective studies have assessed awareness using an interview based on the Brice interview,5 consisting of five questions designed to elicit possible recall of intraoperative events. This questionnaire is administered one or more times postoperatively.2,6,7 It is possible that just administering the questionnaire might produce an increase in reported cases of awareness. However, this seems unlikely given the remarkable consistency of the incidence of awareness in prospective trials and the fact that patients are reluctant to self-report episodes of awareness.8,9 In contrast to the prospective studies, the data contained in the Pollard et al. study3 results from a retrospective analysis of data obtained using a substantial modification of the Brice interview. This interview has been criticized for being designed to reduce the reported incidence of awareness.10 Even though Mashour et al. ’s study4 reported an incidence of awareness three times larger than Pollard et al. ’s study,3 the low rate is still consistent with the limitations of retrospective CQI databases. If we accept the conclusions of the prospective studies that the incidence of unwanted awareness during general anesthesia is approximately 0.1%, we still have 2 critical unanswered questions. First, what percentage of those patients have adverse sequelae postoperatively? Second, how many of these cases are unexpected (i.e., awareness despite the standard of care being met and administration of “normal” anesthetic doses). There are limited data on the frequency of adverse sequelae. Sandin et al.7 found that 11 of 14 patients (79%) who were aware and received neuromuscular blocking drugs had pain, anxiety, or delayed neurotic symptoms that may continue for years.11 There are more data on the fraction that are “expected,” primarily from a series of studies looking at electroencephalogram Bispectrum (BIS) as a measurement of anesthetic adequacy. BIS values over 60 are associated with reported awareness,2,12,13 and maintaining BIS below this threshold has been shown to reduce the incidence of awareness by about 3/4ths in a prospectively randomized observer-blinded study14 and in a prospective study against a historical cohort.12 This is consistent with a recent study by Avidan et al.15 who found in a high risk group (expected incidence of awareness 1.0%) an incidence of awareness when BIS was used of 0.21%. The studies by Mashour et al.4 and Avidan et al.15 shed some light on the issue of how much volatile anesthetic is required to ensure lack of awareness during anesthesia. We know that 0.6 minimum alveolar anesthetic concentration (MAC) volatile anesthetic abolishes implicit memory formation.16 However, these data were obtained in unstimulated patients. Anesthetic requirements change in response to surgical stimulation. Eger and Sonner suggest that “patients adequately anesthetized with a potent inhaled anesthetic (i.e., desflurane, isoflurane or sevoflurane at 0.5 MAC or greater) have an incidence of awareness that is vanishingly small.”17 However, it seems that 1 awareness case (0.05%) in Avidan et al. ’s series15 and 4 cases of awareness reported by Mashour et al.4 had anesthetic concentrations of 0.5 MAC or larger for most of the case. The fact that Mashour et al. found that patients reported “intraoperative awareness” with the same frequency whether general anesthetic was used or not is unexpected. Apart from confirming the low resolution of CQI databases for determining the incidence of awareness, these data tell us that we are probably not communicating expectations of anesthetic state to our patients. We should make clear that conversations can be heard and patients may experience recall during monitored anesthetic care and regional anesthesia techniques. Mashour et al. have eloquently illustrated the pitfalls of using patient reports in a data system to assess the incidence of awareness. The data are of such low resolution that it is impossible to differentiate the incidence of awareness in patients who received general anesthesia or those who did not (monitored anesthesia care). This calls into question the use of CQI databases3 or retrospective analysis of perioperative information systems4 for the assessing of the incidence of awareness, and emphasizes the need for large prospective trials in the field.