How Cortisol+ Works: Stress Score, Inputs & Limits
Understand the Apple Watch signals behind Cortisol+, its wellness score, personal baseline, update timing and the limits of HRV-based stress tracking.
Updated September 9, 2026 · By Cortisol+ Editorial
What does Cortisol+ measure?
Cortisol+ brings wearable stress and recovery signals into one view. It does not measure cortisol. Apple Watch records heart-related and activity data; Cortisol+ reads permitted data through Apple Health and calculates a 0–100 wellness score. The app calls this its cortisol score, but the number is not a blood, saliva or urine hormone concentration.
A lower score reflects less strain according to the app's model. It does not establish that your cortisol is low, normal or healthy. A high score cannot diagnose an endocrine condition.
Which signals contribute to the score?
The current scoring implementation uses the following base weights. They are software design choices, not clinically established contributions to cortisol production. Available data varies by device, region, permissions and wear time.
| Input | Base weight | Interpretation in the app |
|---|---|---|
| Heart rate variability (HRV) | 20% | Compared with a source-appropriate baseline |
| Sleep | 20% | A sleep score derived from available sleep data |
| Resting heart rate | 18% | Difference from baseline |
| Respiratory rate | 9% | Difference from baseline |
| Activity | 8% | Steps and active energy; defaults are used when absent |
| Blood oxygen | 5% | Available oxygen readings |
| Cardio fitness estimate (VO₂ max) | 5% | Available fitness estimate |
| Walking heart rate | 5% | Difference from baseline |
| Heart rate recovery | 5% | Available post-workout recovery |
| Workout recovery | 5% | Time since workout and estimated intensity |
Some measurements appear elsewhere in the app without being separate weighted score factors. For example, the app can read wrist temperature, but it is not one of these ten weighted inputs.
How the calculation works
- Read the available inputs. Health permissions and a connected device determine which records can be used. Missing measurements are not equivalent to normal measurements.
- Compare relevant signals with a baseline. The model starts with population defaults. Its scoring baseline uses an exponentially weighted history from the preceding 30 days, begins blending after 10 stored readings and reaches full personal weighting at 28 readings. That is a reading count, not a guarantee of calibration after seven calendar days.
- Combine factor scores. Smooth scoring curves convert inputs into factor values. Available factors are combined with weights; missing factors are excluded where supported, while activity can use defaults. Freshness adjustments affect some weights.
- Apply model modifiers. The score includes adjustments for recent trends, combinations of elevated factors, time of day and consecutive elevated days. It is then limited to 0–100. A daily curve partly reflects the model's time-of-day assumptions; it is not an observed cortisol curve.
Apple Watch HRV uses SDNN. Other supported data sources can use RMSSD, which is a different statistic. The scoring code separates source-specific HRV history; comparing a number from one method directly with another can be misleading.
How often does it update?
Cortisol+ updates from the data available when analysis runs and when supported background updates arrive. Apple Watch background sampling and Health sync are intermittent. Opening the app or refreshing does not create a fresh hormone measurement, and an updated screen can still contain older sensor readings. Check timestamps and data availability before interpreting a change.
Apple describes how its heart-rate measurements work and how cardio fitness is estimated. Device settings, battery use, movement and sync can affect availability.
What is supported by research, and what is not?
Kim and colleagues' review of stress and HRV supports studying HRV as an indicator of autonomic responses to stress. That evidence does not validate Cortisol+'s formula, prove that HRV predicts a person's cortisol level, or establish diagnostic accuracy.
We have not established clinical accuracy of the app's score against laboratory cortisol testing. Software checks verify aspects of implementation; they do not demonstrate a medical outcome. We do not claim the score can detect Cushing's syndrome, Addison's disease, predict burnout or prove that a supplement is working.
What can affect your reading?
Exercise, illness, alcohol, medication, posture, breathing, recording time, watch fit, missing data and changes in data source can affect the inputs. A change may have several explanations. Look at trends together with how you feel, rather than treating a single score as a diagnosis or a reason to change medication.
A useful way to start
Connect your device, enable only the Health permissions you want to share, and inspect the input timestamps. Use the seven-day stress and sleep journal to note your routine alongside your readings. The journal helps you ask better questions; an uncontrolled week of tracking cannot prove cause and effect.
Read our Apple Watch cortisol and stress guide, cortisol testing guide and privacy policy. Core scoring runs on-device; optional account, connected-service and coaching features have their own data flows described in the policy.