In this work, we present some of text search algorithms, two models of Arabic dictionary, new search algorithm depends on binary search algorithm, and numeric example to declare how the new searching algorithm working. In The first model the Arabic dictionary will be divided into several parts. That equal to the number of alphabets (characters). Each partition has all words, which start with an alphabet. This method needs a zero-comparisons to acquire a group of words started with chosen alphabet The words (elements) in the dictionary are sorted in ascending order according to the _ code of the alphabet sorting. The disadvantage of this model is the insertion or deletion of an item (to or from) the array. This process requires in averagereplacement for every insertion or deletion. The main advantage of the second model is to reduce the number of replacement either at insertion or deletion fromto replacements in average for every insertion or deletion. In the second model, we will divide the Arabic dictionary into 35 arrays each of one-dimension. Each array has all words, which start with an alphabet. In all parts of Arabic dictionary, we assume that the number of words in each part are equal. That means the number of words in each part equal to.The new searching method is very effective when the number of words (elements), where m is an integer number. This means that the method is effective when .