One thing that can confidently be predicted is that before two continents can collide, they must first have an intervening ocean between them. There must also be one trench, into which the ocean floor between them is subducted. Should such a a trench initially form in the middle of an ocean between two continent-bearing plates, the ocean floor between the trench and one continent will be subducted first. When all the ocean floor on the subducting plate is consumed, and the trench reaches the continental margin, the trench will collapse. Any sediments accumulated within the trench will be uplifted as the formerly depressed section of ocean floor rebounds. The Coast Range of California was created by just such an uplift.
Continued convergence between the two plates will cause the formation of a new trench, most likely forming just offshore from the former one but reversed in direction, as the now active continental margin overrides the remainder of the intervening ocean floor. Therefore in all continent to continent collisions, at the time of contact one continent must have a passive margin, the other an active one. There is no evidence that a subduction zone can ever form within the interior of a continent (a subduction zone is defined by the subduction of an oceanic plate). There are, however, several subduction zones in the South Pacific that are a considerable distance from any continental margin.
It can also be reasonably predicted that when two continents collide, not all adjacent points will 'hit' at the same time. Depending on the relative shapes of the two approaching continental margins and the rate of convergence, orogeny may commence in one region tens of millions of years before it occurs elsewhere. Where island arcs exist on the active margin, minor orogeny may begin when an island contacts the passive mainland, while the major 'crunch' doesn't occur until millions of years later. The Taconic Orogeny, which produced the mountains of the same name in eastern New York state, USA, were an island arc collision around 450 to 500 MYa (million years ago). The major crunch between Baltica (northern Europe) and Laurentia (ancestral North America) didn't occur until around 400 MYa.
The upper folds of folded mountain ranges remain poorly consolidated sediments, and erode away rather quickly. River courses tend to follow the orientation of the folds, cutting their way deeper and deeper into the soft material. The roots, on the other hand, are depressed so deeply into the mantle that the increase in pressure and temperature partially melts the sedementary material and causes it to flow much like toothpaste squeezed from a tube. As the tops of the mountains are eroded away, the buoyancy of the roots causes them to be uplifted, until the partially remelted material, now cooled and recrystalized as metamorphic rocks (granitite, schist, and gneiss) are exposed at the surface. Much of this later uplift occurs as upthrust faulting, and is frequently accompanied by intrusions of fully molten magmatic or granitic material into the cracks and crevasses. (The Appallachian Mountains of the Eastern USA, now a mere ghost of their former glory, illustrate nicely this stage of evolution.) Given enough time, the mountains will be fully eroded away, leaving an essentially flat, peneplained surface of metamorphic rock. Such surfaces, called cratons, form the cores of many of our present continents. (The Canadian and Baltic Shields are prime examples.)
One possible scenario for the crustal shortening of a craton is overthrust faulting. It has been reported that when the craton of India slammed into the southern coast of Asia, a large wedge of cratonic material was sheared off the leading (northern) edge of the Indian subcontinent and thrust up and southward over the Indian craton. The thickest (northern) edge of this wedge is said to make up the south face of the Himalayas. A similar wedge has been reported to have thrust eastward across Wyoming and into the Black Hills of South Dakota.
A second scenario, likely to occur when two cratons collide, was reported from echo soundings in one section of the southern Appalachians. It appeared that both continents (Gondwanaland and ancestral North America) sustained multiple shallow dipping fractures, allowing fingers of cratonic material to interpenetrate, much like shuffling a deck of cards.
A typical active margin will also likely include odd erratics, bits of debris ranging from sea floor sediments to ancient seamounts and small islands, that were scraped off of the subducting sea floor. Intrusive magmas, deposited during the rifting process, may or may not be present. Deep sea sediments uplifted by the collapse of the intervening trench will likely also be found. Island arcs may end up becoming compressed into the mess as well. Each of these pieces is likely to behave differently under the enormous pressures that occur in a continental collision, and each adds its own bit of complexity to the resultant structure. It is small wonder that the underlying processes are rarely visible and are exceedingly difficult to interpret. In order to examine a forest, one has to look at a lot of trees.