Collocations are a common phenomenon in natural languages in general and in Arabic in specific. Quran is widely considered to be the most eloquent and authentic Arabic text. This fact should motivate researchers to explore and study the collocations of this miraculous book. One approach to find collocations for a given corpus is by identifying pairs of words that are more likely to exist adjacent than being separate using special statistical measures. Chi-squared, t and mutual information are three broadly accepted statistical measures that are used for this purpose. In this paper, these three statistical measures are used to extract the collocations of Quran. None of these measures gave perfect results. So, a human interaction is required to filter the best candidates from the found collocations. The t measure gave the best set of collocations. The other two showed bias towards pairs of adjacent words that are unique or exist with very low frequencies in Quran. To give a generic notion of Quranic collocations that are extracted using this approach, those that were found by the t measure are studied and categorized.