Flaccid and erect are loosely related

Flaccid length predicts erect length only weakly: across measured men the correlation is positive, but it is too loose to tell any one man his erect length from his flaccid one. A population correlation and a rule for an individual are not the same claim, and stretched length is the better proxy of the two.

What the studies report, directionally

Measured studies that report both flaccid and erect length on the same men find a positive association - men with longer flaccid length tend, on average, to have longer erect length too. That association is real in the statistical sense: it shows up consistently enough across studies to be treated as an established finding rather than noise. It is also weak enough that it explains only part of the difference in erect length between two men, leaving a large share unaccounted for by flaccid length alone. The classic direct test is Wessells, Lue and McAninch (1996), Journal of Urology, who measured 80 men flaccid and then after a pharmacologically induced erection. Their mean flaccid length was 8.8 cm and erect length 12.9 cm, and they reported that "neither patient age nor size of the flaccid penis accurately predicted erectile length."

Folk shorthand has a name for the two ends of this relationship - "growers" and "showers" - and it is worth naming only to be clear that it is folk terminology describing the same underlying weak correlation, not a separate finding, a formal category, or a binary that every man falls cleanly into.

Why it is weak rather than strong

The mechanical reason is that flaccid length is a poor, noisy signal to begin with. It moves with temperature, arousal state, time since last activity, and time of day, so a single flaccid reading is measuring "flaccid length right now, under whatever conditions happened to apply" more than it is measuring a stable trait. Two readings of the same flaccid man, twenty minutes apart, can differ meaningfully for reasons that have nothing to do with what his erect length will turn out to be. A predictor that is itself unstable cannot be a strong predictor of anything else, which is the main reason the flaccid-to-erect correlation sits where it does rather than higher.

Stretched flaccid does better

The literature's alternative, stretched flaccid length, correlates with erect length more strongly than resting flaccid length does - in the same 1996 study, "stretched length most closely correlated with erect length" - which is the practical reason clinics favour it as a proxy when an erect measurement is not obtained. Even so, it runs short: in 201 men measured by seven andrology specialists, Habous et al. (2015), Journal of Sexual Medicine, found stretched length under-read erect length by a mean 2.64 cm (21.4%). Pulling the shaft to resistance removes much of the state-dependent slack that makes an unstretched flaccid reading unstable, though stretched length carries its own separate source of variation, since the tension applied is operator-dependent rather than fixed.

What not to do with any of this

Do not use a flaccid reading, however carefully taken, to predict an individual's own erect length with any precision - the population-level correlation does not license that inference, and the honest position is that a flaccid measurement mostly tells you a flaccid measurement. Do not treat "grower" or "shower" as a diagnosis; it is a loose description of where someone happens to sit on a weak, continuous relationship, not a category with a defined boundary.

This question is about two length measurements taken at two states with a ruler, which is a narrower thing than a subjective rating of appearance, and narrower still than an estimate an image model produces from a single photograph, which has no access to a second, differently-stated reading to correlate against at all. It is also a different question from how a human judge's score might change between a flaccid and an erect photo, which is about perception rather than a measured statistical relationship, and from how a platform's scoring conventions handle state, which is a policy question rather than a finding from the literature.

If you want the mechanical explanation for why flaccid length itself is so unstable in the first place, that is covered separately; read alongside what the pooled clinician-measured data actually report for both states, this is enough to place a flaccid reading honestly - as a rough signal, not a forecast.