To assess whether a true knowledge of crowding alters treatment decisions compared with estimates of crowding. Thirty-six orthodontists were asked to estimate crowding using visualization on eight mandibular arch study models and to indicate possible extraction choices. For each model, the intermolar widths, intercanine widths, and clinical scenarios were identical, but the true crowding varied from 0.2 to 8.4mm as to a lesser extent did the curve of Spee. Eleven orthodontists repeated the visualization exercise after 2 weeks to assess reliability. All 36 of the orthodontists were asked to repeat the treatment planning exercise on the same models, but this time was provided with the true amount of crowding in each case. When the 36 orthodontists used direct visualization of the models to assess crowding, the range of their estimates of crowding increased as the crowding increased. As might be expected, they also tended to move towards extraction treatments as the crowding increased (P = 0.013, odds ratio = 3). Although the reliability of the repeat estimates of crowding were moderate, the mean estimates were greater than the true crowding for each model. When orthodontists were presented with the true amount of crowding, rather than their estimate of crowding, it had a significant effect on the decision to extract, with fewer orthodontists recommending extractions. The principal limitation of this study is that it was a laboratory-based study and utilized just the mandibular arch model for estimation and treatment planning. Direct visualization may overestimate the amount of crowding present. When the true amount of crowding is known, it can lead to more consistent treatment planning, with the decision to extract fewer teeth in the borderline cases. A formal space analysis is likely to assist with treatment planning.