Problems with the argument from fulfilled prophecies
There is an argument which is often used by evangelical Christians: that the Old Testament of the Bible contained prophecies, "predictions", which have been fulfilled later. It were the proof that the Bible has a divine inspiration, that it is "God's word", because humans could not make such predictions of future events. There are two kinds of Bible "prophecies": "messianic" prophecies (which are supposed to predict Jesus' coming and the details of his life) and "non-messianic" prophecies (about all other issues). In this article, I first list the criteria which are to be respected when we want to know if a prophecy in a book is a proof of its supernatural origin or not. I then mention the intellectual errors which are propagated in the case of the Bible in order to come to the conclusion that this book contains fulfilled prophecies which prove its divine inspiration and even the dishonesty of some apologists. I finally give two examples of non-messianic prophecies (destruction of Tyre and Egypt in Ezekiel) which are triumphantly mentioned by some apologists. For examples of messianic prophecies, see this article.
Criteria to know if a prophecy is a proof for the divine inspiration of a book
Of course, we first have to make sure that the prophecy was written before the predicted events. If we are not sure, we can't know if the prophecy is a fraud or not (prophecy written after its "fulfillment"), and such a prophecy proves nothing. We should particularly have this point in mind for example for Daniel's prophecies. It is important to understand that the burden of proof is on the side of those who pretend that there is a "miraculously fulfilled prophecy". If there is any other explanation than the miracle, then the proof for a divine inspiration is refuted.
We then must make sure that the predicted event has indeed taken place. An author can also write in a book (which is later considered as divine) that any event took place, although it is not true or the author does not know if it is true or not, just because prophecies announce it and it was theologically necessary to pretend that this event had indeed taken place. We should particularly have this point in mind for example for prophecies to less known parts of Jesus' life in his biography (birth and period following).
We must make sure that it was not possible for some people to act to fulfill themselves the prophecy, or to make its fulfillment more probable. We have to make sure that the prophecy is formulated in a enough clear way. If it is not the case, the room for interpretation is so large that a a posteriori reinterpretation, which allows to conclude that the prophecy has been fulfilled, is always possible. We have to make sure that the prophecy had not a high probability to be fulfilled if we waited long enough. By the way, there are apologetic methods which allow to reduce nearly all prophecies to one of the two last cases (even if they are clearly formulated and had a very small probability to be fulfilled by chance).
Bible apologists often use the "prophecy splitting method": it is about splitting prophecies (or any text of the Bible) afterwards and arbitrarily (sometimes within a sentence, by a comma) in order to obtain many new prophecies which don't have anything to do with the author's idea anymore. These small prophecies have no context, they are now unclear and have generally a fulfillment probability of 1 if we wait long enough (several centuries or millenniums), and in the worst case, it is always possible to say that a prophecy has "not been fulfilled yet" (so the Bible can only be either "proven" or "not refuted" with this biased method). In general, these small prophecies deal with destructions, massacres, terrible bloodbaths (the god of the Old Testament focuses particularly on peace and love). Each city, each civilization, is destroyed one day, if we wait long enough. The worst tragedies happen one day. This was particularly true during Antiquity, when conquests were commonplace. So, it is possible to see plenty "fulfilled prophecies" where we had, in fact, a failed prophecy.
Josh McDowell and Werner Gitt suppose that all of these small prophecies have a probability of fulfillment by chance of 1/2. The probability that a complete prophecy is fulfilled by chance is then equal to 1/2 to the power of the number of small prophecies which the complete prophecy contains. We obtain a very small number (for example 0.0000001) which is then presented to be too low in order to be explained naturally (so the explanation of a divine intervention). Then, the very small probabilities of all complete prophecies are multiplied together in order to obtain the probability that all Bible prophecies were fulfilled by chance. Each time this dishonest method does not allow to link one of the small prophecies to an event which occurred within the 2000 last years (or much more), the "prophecy" is just seen as "not fulfilled yet" and is simply not taken into account in the probability calculation, so that there is no consequence for the Bible.
It is tragic that these apologists reach very much more people than people who refute them. At evangelization stands (where books from these apologists are distributed for free), I hear things like: "Take this book. The author, Josh McDowell, used to be atheist, he made a long research in order to refute the Bible, but this led him to the other conclusion, that the Bible could only be God's word." Josh McDowell, the apologist who blames the Internet for bringing the arguments of atheists and skeptics to children of Christian families, where these children should, after him, only hear what their parents tell them to this issue, as it used to be the case: see this article.
Here are two examples of non-messianic prophecies for illustration:
Ezechiel's prophecy to the conquest of Egypt by Nebuchadnezzar
Ezechiel's prophecy to the destruction of Tyre
More refutation of Josh McDowell on infidels.org.