Many architects report about mentally visualizing 3D aspects of their design ideas while simply working with 2D sketches of them. Indeed, in architecture, the general practice of conveying 3D building information by means of 2D drawings bears on the assumptions that every architect can mentally visualize a building in 3D by looking at its 2D drawings or sketches and that architects, as many report, can capture the 3D aspects of a building design during such 3D mental visualization practices. Additionally, many intuitively believe that the levels at which architects perform such 3D mental visualization practices is highly correlated to their spatial visualization abilities as defined by existing measures of spatial visualization ability. This thesis presents the outcomes of protocol studies and analyses that were conducted with the aim of developing an in-depth understanding about such 3D mental visualization practices and capabilities of architects on the basis of four research questions. First, what might be the nature of the 3D mental visualization phenomena that architects claim to experience: what are the features of these 3D mental visualizations as evidenced in specific tasks; and what might be the nature of the mental representations created during these visualization processes? Second, can every architect carry out these 3D mental visualization practices; might there be individual differences among architects' performances? Third, might 3D mental visualization of buildings be only an architectural skill; can non-architects, who can read 2D architectural drawings, visualize a building in 3D based on its 2D drawings and can they do so to the same levels of performance of those of architects? Fourth, might performance in 3D mental visualization tasks be related to/predicted by spatial visualization ability? The major conclusions of this thesis with regard to the first research question include that (1) architects can be visualizing the buildings in one of the two major forms or by alternatively switching between them: by imagining themselves situated within (almost) the actual size 3D building environment or by imagining a 3D small scale model of the building; (2) the mental representations they create during these visualization processes capture the various visual and spatial aspects of the buildings with a structure similar to that of an actual size or small scale model of the visualized space/form, yet the way they capture these aspects is not like the way these aspects would be captured from a certain viewpoint in reality; and (3) what they experience during these visualization processes is not like the continuous holistic visuospatial experience that one would have when looking at a building or walking inside/around a building. With regard to the second, third and fourth research questions this thesis concludes that (question 2) architects differ in their 3D mental visualization skills; (question 3) 3D mental visualization is an architectural skill in that it relies on certain abilities that become heightened in architects, possibly during education; and (question 4) 3D mental visualization skills are not related to spatial visualization ability as defined by the standard paper-folding test of spatial visualization ability.