A researcher asks a participant to collect a saliva sample the moment they wake, then again after a carefully timed interval. The participant reaches for the tube, but also reaches for the phone, gets out of bed and begins thinking about the day. A few minutes of delay can alter what the samples appear to say. The method is simple to describe and unusually demanding to perform well.
The cortisol awakening response, or CAR, is the characteristic rise in cortisol across roughly the first 30 to 45 minutes after morning awakening. It is studied as a distinct part of daily cortisol dynamics. It is not a single morning concentration, a direct measure of willpower, or a diagnostic “stress score.” Understanding it requires attention to both biological timing and the exact moment a person woke and sampled.
The rise after waking
Cortisol follows a daily pattern and is released in pulses. Levels typically change before waking and through the day. The CAR is a sharp rise following awakening, superimposed on that broader rhythm. Fries, Dettenborn and Kirschbaum described it as a distinctive HPA-axis feature rather than treating every morning increase as one undifferentiated event.
Research reviewed by Stalder and colleagues in Endocrine Reviews has sharpened the distinction between the rise before waking and the response after waking. The CAR’s regulation and function remain subjects of active investigation. It may relate to preparing for the day, but one should not turn that broad idea into a claim that an individual sample reveals exactly what a person expects or feels.
Wilhelm and colleagues used sleep-laboratory work to examine whether the rise is a response to awakening rather than merely a continuation of the pre-awakening trajectory. Their work supports the importance of measuring actual awakening, not just assuming that a planned clock time marks the biological event. A study that gets this timing wrong can misclassify the curve.
What is actually sampled?
Many CAR studies use saliva, collected at waking and at several points afterward. Salivary cortisol offers a practical way to sample repeated values without repeated blood draws. A protocol may include measurements immediately after waking and around 30 or 45 minutes later, sometimes with additional points. The exact schedule should be specified; two samples and five samples do not give identical information.
The first sample must be close to actual awakening. If it is delayed, the “baseline” may already include part of the rise. A participant who writes down a planned time but collects the tube after a shower can produce an apparently flatter or otherwise distorted curve. Stalder and colleagues’ 2016 consensus guidelines put objective verification of awakening and sampling times at the center of valid CAR assessment.
The 2022 update found continuing methodological issues in published work and renewed the emphasis on adherence. The point is not to blame participants for ordinary mornings. It is to design a protocol that recognizes how difficult precise home sampling can be, supports it, records deviations and analyzes them transparently.
Several measures, several questions
A researcher can report the cortisol concentration at awakening, the increase from waking to a later point, or an area-under-the-curve summary of several samples. These are related but not interchangeable. Pruessner and colleagues distinguished formulas that summarize total hormonal output from formulas emphasizing change over time. A study must name which result it uses before attaching a psychological interpretation.
For example, two people could show similar rises while beginning at different waking concentrations. Their total cortisol exposure over the sampled interval would then differ. Another pair could have similar total output but different shapes of rise. A single label, “high CAR,” hides these distinctions unless the analysis defines the measure.
The choice of measure should match the question. If the question is the magnitude of the post-awakening rise, an index that mixes baseline and response may be less direct. If the question is total early-morning output, change alone may omit relevant information. Both can be reported, with care about multiple comparisons and the temptation to highlight only the most dramatic result.
Timing is a source of bias
CAR values can vary with time of waking, sleep and circadian phase. Shift workers and people with irregular schedules may not fit an uncomplicated “morning” assumption. Light exposure, medications, health conditions and day-to-day activities can also matter. A study must describe its sample and protocol before treating different curves as evidence of different stress histories.
The 2016 and 2022 expert guidance identifies adherence as more than random noise. If delayed sampling is more likely among people with particular schedules or burdens, error may correlate with the very variable being studied. Then the comparison can be biased, not merely less precise. Electronic time monitoring and clear procedures can help, while privacy and feasibility must still be considered.
Even a well-timed curve from one morning is only one observation. Daily circumstances vary. A person may wake unusually early, be ill, have an interrupted night or anticipate a difficult event. Repeated days can help separate a more stable pattern from day-specific variation, though they also increase participant burden and require careful handling of missing samples.
CAR is not all of cortisol regulation
The rise after waking is one segment of a full daily profile. A study might also investigate the decline over the day, responses to a laboratory challenge or longer-term measures such as hair cortisol. Those outcomes reflect different time scales and mechanisms. One cannot infer the whole HPA axis from the CAR alone.
The distinction matters when communicating about chronic stress. A headline may say that stressed people have “too much cortisol,” yet different studies may measure waking level, rise, total output, evening concentration or response to an acute task. They can disagree without one necessarily being wrong. The measures do not represent the same question.
Russell and Lightman’s review of the human stress response emphasizes pulsatile and circadian regulation. An isolated value ignores that dynamic context. The CAR can be informative when used as a specified research measure, but it cannot be converted into a consumer verdict about a person’s health or readiness.
What association can and cannot mean
Researchers have examined associations between the CAR and psychological conditions, anticipated demands, sleep, health and social context. Results depend on design and measurement. An association does not establish that an appraisal caused a particular rise, that the rise caused a later condition, or that changing one morning’s response would improve health.
Some theoretical accounts propose that the CAR helps prepare a person for the upcoming day. That is a useful hypothesis, but the function is not settled as one universal mechanism. The comprehensive review by Stalder and colleagues discusses regulatory and functional evidence while leaving room for uncertainty. A reader should distinguish a measured curve from an explanation of why it occurred.
A study of groups also cannot diagnose an individual. Even if two populations differ on average, their values may overlap widely. Sampling error and context can be large relative to the difference. Treating a one-day CAR as a personal ranking of resilience or vulnerability exceeds the evidence.
Why objective adherence matters
Self-recorded wake and sample times are vulnerable to ordinary memory and routine. A participant may honestly report collecting “on waking” when several minutes passed. Some protocols use electronic caps or other time records, and studies can compare these records with diaries. They must also state what counts as a valid day and how nonadherent days are handled.
Simply discarding imperfect days can introduce selection bias if the people most burdened by a study have the hardest time following the schedule. Including all days without accounting for delay can bias the curve in another way. The best analysis plans make these choices in advance, report the amount and pattern of missing or late data, and test whether conclusions depend on reasonable alternatives.
Participant instructions matter. A protocol should say whether eating, drinking, tooth brushing, medication or movement before the first sample is allowed. It should use language people can follow on waking. The study should not place a complex cognitive task at the moment when exact timing is most critical and then quietly treat deviations as personal failures.
What a stronger CAR study reports
It identifies the number and timing of samples relative to verified awakening. It reports how samples were collected, stored and analyzed, and whether any relevant medication or conditions were considered. It distinguishes planned from actual sampling times and states how late samples were treated. The 2016 consensus and 2022 update provide a useful checklist for this methodological transparency.
It then reports the particular CAR index, with enough information to reconstruct the result. Did the authors analyze awakening concentration, absolute increase, percentage change, area under the curve with respect to ground, or area with respect to increase? If more than one was tested, readers need the full pattern rather than one selected association.
Finally, it keeps interpretation proportional. A well-controlled group difference can motivate further study. It does not automatically reveal a single hormone-based cause of stress, depression or work performance. The stronger the claimed application, the more evidence is needed that the association survives repeated sampling and predicts a meaningful outcome.
A measure that can change the morning
Collecting saliva several times immediately after waking is not an invisible act. It can alter a participant’s routine, attention and perhaps anticipation. People with children, care duties or shift schedules face different practical burdens. Those differences may affect adherence and the experience of the study itself.
An equitable design considers whose mornings are easy to standardize and whose are not. It may need flexible scheduling, clear support and explicit reporting of who was excluded. If a method works only for participants with quiet, predictable mornings, its conclusions should not be generalized to all workers or families without caution.
The Institute’s synthesis is that physiological data should be accompanied by a story of how it was obtained. This is a transparency principle, not a claim that the CAR can be turned into a daily coaching device. A number without a trustworthy timeline of waking and sampling is not self-explanatory.
Do not collapse the neighbouring constructs
The CAR is not the same as the general stress response, chronic stress or recovery from stress. A person can face persistent pressure while showing a CAR pattern that does not match a simple public narrative. A recovery intervention may change sleep or subjective strain without changing this one endocrine index. The constructs need separate evidence.
It is also different from the cortisol response to an acute social-evaluative task. Both involve the HPA axis, but their triggers, timing and protocol differ. A paper that uses one to explain the other should show the connection empirically. Shared hormone does not mean shared meaning.
Keeping these boundaries clear makes the method more valuable. It allows researchers to ask whether post-awakening regulation offers information that a general daily level does not, and whether that information remains stable when measurement quality improves.
Research reports should show the actual sample schedule, not only a summary index. A first sample collected after a delay can miss the level at awakening and distort the apparent rise. Electronic monitoring can improve confidence about timing, but it does not remove every source of variation. Sleep timing, waking method, shift work, medication and sample handling may matter. Repeating the protocol on more than one day can help distinguish a stable pattern from a day-specific observation, while still leaving interpretation dependent on the research question.
Two studies can both say they measured the CAR and still calculate different outcomes. One may compare a later sample with the awakening value; another may use area under the curve relative to ground or relative to increase. Pruessner and colleagues explain why those summaries represent different aspects of hormone output. A reader should not assume that a larger total area means a larger awakening rise. Reporting the raw time points alongside the chosen summary makes the distinction inspectable. It also lets later reviewers see whether a surprising group difference arose from the awakening sample, the subsequent rise, or both.
What the evidence does not show
The CAR does not show how stressed, motivated or resilient a person is from one morning’s samples. A larger rise is not universally “good” or “bad.” A flat-looking curve may reflect late first sampling. A measured association with a condition does not establish a causal pathway or a personal diagnosis.
The evidence also does not support treating the CAR as interchangeable with every other cortisol measure. The post-awakening rise is a time-specific phenomenon, and valid interpretation requires reliable wake and sample times. Its limitations do not make it worthless; they identify the conditions under which it becomes useful evidence.
The first minutes of the day
The participant wakes, finds the tube and starts the timer. The scientific question lives in those ordinary actions as much as in the laboratory assay. The rise may reveal something about post-awakening regulation, but only if the samples actually mark the moments the analysis says they mark.
The cortisol awakening response is a reminder that bodily data are events in time, not free-floating scores. Careful timing, a clear measure and restrained interpretation make the difference between a meaningful physiological study and a confident story attached to an uncertain morning.
Research references
- Fries, E., Dettenborn, L., & Kirschbaum, C. (2009). The cortisol awakening response (CAR): Facts and future directions. International Journal of Psychophysiology, 72(1), 67–73. View source
- Stalder, T., Kirschbaum, C., Kudielka, B. M., Adam, E. K., Pruessner, J. C., Wüst, S., et al. (2016). Assessment of the cortisol awakening response: Expert consensus guidelines. Psychoneuroendocrinology, 63, 414–432. View source
- Stalder, T., Lupien, S. J., Kudielka, B. M., Adam, E. K., Pruessner, J. C., Wüst, S., et al. (2022). Evaluation and update of the expert consensus guidelines for the assessment of the cortisol awakening response (CAR). Psychoneuroendocrinology, 146, 105946. View source
- Stalder, T., Oster, H., Abelson, J. L., Huthsteiner, K., Klucken, T., & Clow, A. (2025). The cortisol awakening response: Regulation and functional significance. Endocrine Reviews, 46(1), 43–59. View source
- Wilhelm, I., Born, J., Kudielka, B. M., Schlotz, W., & Wüst, S. (2007). Is the cortisol awakening rise a response to awakening? Psychoneuroendocrinology, 32(4), 358–366. View source
- Pruessner, J. C., Kirschbaum, C., Meinlschmid, G., & Hellhammer, D. H. (2003). Two formulas for computation of the area under the curve represent measures of total hormone concentration versus time-dependent change. Psychoneuroendocrinology, 28(7), 916–931. View source
- Russell, G., & Lightman, S. (2019). The human stress response. Nature Reviews Endocrinology, 15, 525–534. View source