Not to disagree with or contradict, but to add to, Ragi's answer:
The distinction between 2D, 2.5D and 3D
Generally, a GIS holds (at least) 2D features on 2D maps.
That is, features are geo-located in two primary geographic dimensions: X & Y.
Depending on the context, we call them northings & eastings or latitudes & longitudes.
The features are represented by points, lines and polygons, the elements of which are X-Y data pairs.
To be more useful, a GIS will hold geographic surfaces or even features sitting on such surfaces.
The obvious case is the Earth's surface but it could be more abstract "surfaces" like local population density or local annual days of sunshine.
There are the two primary geographic dimensions, X & Y, and a third dimension, Z.
Such features are again represented by points, lines and polygons, but the elements of which are now X-Y-Z data triplets.
So is it 3D? Yes and no.
A distinguishing characteristic of a geographic surface is that, while it can exist everywhere in 2D X-Y space, it has only a single Z value at any given 2D location.
Even more useful is a system that holds geographic volumes.
These are "true" 3D features existing in 3D spaces and can be enclosed by surfaces on all sides.
Think sophisticated geological, oceanographic or meteorological models.
Or multi-storey building or complex industrial plant models.
They are represented by points, lines, polygons (as above) and polyhedrons.
And as above, the elements are still X-Y-Z data triplets.
However, a distinguishing characteristic of a geographic volume is that it can exist anywhere in 3D X-Y-Z space.
And at any given 2D location there can be multiple Z values.
So what to call the middle type of data if it's more than 2D but less than true 3D?