Magnetic resonance imaging (MRI) studies on centric condyle positions lack 3D comparisons of guided and unguided methods, which are used for accomplishing centric relation reference positions. The purpose of this study was to describe the space, in which mandibular condyles are placed in vivo by dental intercuspation, Dawson's bimanual manipulation, and neuromuscular position. Twenty asymptomatic individuals aged 23 to 37 years underwent separate bite registrations using bimanual manipulation and the unguided neuromuscular technique. Subsequent 3-Tesla MRI scans of both temporomandibular joints yielded 3D data of the most superior condylar points at maximum intercuspation and both centric relation positions. We found concentric condyle positions in maximum intercuspation but considerable variation of condyle position after bimanual manipulation and neuromuscular technique. Their 95% predictive confidence ellipses overlapped substantially and created a space of reference positions. Its smallest volume averaged 2 mm3 for a minimal convex hull (95% confidence interval 1.1-3.2) and 3.5 mm3 for a minimal ellipsoid hull (95% confidence interval 1.8-5.4). Visualized in vivo by MRI, condyle positions at bimanual manipulation and neuromuscular position were not predictable and showed substantial variation in asymptomatic subjects. Clinicians should be aware of the functional space and its effect on dental intercuspation.