SVG documents are scalable. The conventional way to position SVG documents, which comes from the web platform, is to consider a viewport in which to place the SVG document — that is, a rectangular region to where the SVG will be scaled and positioned.

SVG renderers are supposed to use the viewport provided by the application, plus the SVG document's width and height attributes, and its viewBox attribute, to compute the position and size for the rendered document.

Ideally, the toplevel <svg> element of an SVG document will contain width and height attributes, that indicate the proportions and "natural size" of the document. When those attributes are present, the SVG renderer can unambiguously figure out the natural aspect ratio of the document, and can also suggest a natural size for the document. Since SVGs are scalable, it is not mandatory to actually use its natural size; it can be scaled arbitrarily. Of course, it is up to each application how an SVG document will be scaled: a web browser would want to consider the semantics of embedding images in HTML, which may be different from a GUI toolkit loading SVG assets with hard-coded sizes.

If an SVG document's toplevel <svg> element does not have width and height attributes, then the SVG renderer can try to figure out the document's aspect ratio from the viewBox attribute. If there is no viewBox either, then the SVG renderer cannot easily figure out the natural size of the document. It can either set a 1:1 scaling matrix within the application's viewport and render the SVG there, or it can actually try to compute the size of each object in the SVG document to figure out the size. The latter is a moderately expensive operation, and can be avoided by having the SVG document specify width and height attributes.