Able-bodied people from the two extremes of the distribution of passive dorsiflexion exhibit biomechanical differences when descending stairs. Whether physical therapists could judge accurately the amount of ankle flexibility from observation of stair descent was examined with physical therapists (20 neurological 20 cardiothoracic) and 20 students. Subjects were 18 able-bodied men with high or low passive dorsiflexion. Raters viewed videotaped subjects descending stairs at normal playback speed and in slow motion and rated ankle flexibility on a 20-point scale. 'Flexible' or 'inflexible' subjects were classified correctly better than chance, but accuracy of rating flexibility was relatively poor. Both experience and slow motion playback improved accuracy. Physical therapists can detect extremes of normal flexibility in the ankle, based solely on observation of stair descent, and experienced therapists are better at this task than less experienced ones. Using slow motion videotape enhanced observational analysis.