Method
Stretched versus erect
The literature treats stretched flaccid length as an approximation of erect length; here is what that relationship is and where it breaks down.
Guides on Method: Bone-pressed length, start to finish, Stretched flaccid length, and why the clinics take it, Girth, done the way the studies do it, The measurement glossary
Stretched flaccid length is used as a stand-in for erect length in clinical settings, and it is worth being precise about what "stand-in" means here, because the word gets stretched further than the tissue does.
Why it was adopted at all
An erection is hard to standardise on demand in a clinic - it requires either pharmacological inducement or private self-stimulation, both of which raise practical and ethical friction that a flaccid measurement does not. Wessells and colleagues (1996) measured men flaccid, stretched and erect in the same session and found stretched length tracked erect length more closely than flaccid length did, which is the finding that made stretched length a usable proxy: something a clinician can obtain quickly, without inducing an erection, that carries useful information about the number they actually want.
What "approximates" means, and what it does not
This is a population-level relationship, not a promise about any individual. Across a large group of men, stretched and erect lengths track each other well enough that stretched length is treated as informative; Veale et al. (2015) in BJU International reports both measures and the relationship between them holds up at that scale. "Tracks well at the population level" is a statement about a correlation across many people, not a formula that reproduces one person's erect length from their stretched length. Two men with identical stretched readings can have somewhat different erect readings, and the gap between an individual's own stretched and erect numbers is not something you can predict from the population relationship alone.
Where it breaks down
The correlation is real but not perfect, and the reasons it loosens are physiological rather than a flaw in the measurement itself: how much a given man's tissue extends under manual tension does not map onto how much the same tissue expands under vascular engorgement in a fixed one-to-one way for every person. Some men stretch further relative to their erect length than others do, and some stretch less; the proxy works well in aggregate precisely because these individual departures tend to sit on both sides of the average rather than piling up in one direction. This is also why stretched length is reported alongside erect length rather than instead of it wherever a study collects both - it is a useful adjunct measurement, not a replacement.
Why clinics still bother collecting it
If stretched length only approximates erect length imperfectly, it is worth asking why studies keep collecting it at all rather than simply inducing an erection every time. The practical answer is friction: pharmacological inducement carries side effects and requires clinical oversight, and self-stimulation in a clinical setting is not something every study design can accommodate or every participant will consent to, which shrinks the sample and skews who ends up measured. Stretched length sidesteps both problems, which is why it shows up as a secondary measure in datasets that also report erect length, including Veale et al. (2015), rather than as a replacement for it. Collecting both lets a study report the erect figure as its primary result while also contributing to the evidence base on how well the stretched proxy performs, which is part of how the reliability of the proxy itself has been checked across more than one dataset since Wessells and colleagues (1996) first proposed it.
What a large gap on one person might suggest
Because the correlation is a population-level pattern rather than an individual guarantee, a stretched reading that sits unusually far from someone's own erect reading is not itself informative about anything beyond that person - it does not indicate a measurement error, a health concern, or an unusual anatomy on its own. It mainly says that this particular person is one of the ones the aggregate correlation does not describe well, which the correlation always predicted would happen for some fraction of any sample. Chasing an explanation for why one individual's gap is larger than the average is generally not productive, because the physiological reasons tissue extension and vascular engorgement diverge are not fully mapped at the individual level in the literature this site draws on. The more useful use of a stretched reading stays close to what it was designed for: a same-day approximation, taken with consistent technique, tracked alongside - not instead of - whatever erect data a person also has.
What this means for a single reading
If you have taken a careful stretched-length reading and want to know what it implies about your erect length, the honest answer is: it is directionally informative and not a substitute for actually measuring erect. Treat it as one data point among several rather than a calculated stand-in. The technique for taking the stretched reading itself - how far to pull, what resistance means - is covered separately at how far to stretch: the resistance point in stretched length, and the full stretched protocol is at stretched flaccid length in full; this post is about what the number means once you have it, not how to take it. How much the stretched figure itself varies from one published study to another is a related but different question, covered at why stretched-length figures vary between studies.
None of this is a question a photograph or a score can answer either way. An image-based estimate from something like AI Penis is not measuring tissue extension at all, and a rating from Penis Rater or a judgement from Rate Penis is scoring an appearance rather than tracking a physiological relationship - the stretched-to-erect proxy is a finding about bodies under tension, and Rate Cock's scores are a finding about photographs, which is a different kind of number built for a different question.