Information gain is a very important parameter in the process of decision tree construction, which is a classifier tool in Data mining.  With the explosive growth in data collection in business and scientific fields, the traditional data mining algorithms working on very large data sets take very long times on conventional computers to get result.  Although several parallel data mining algorithms have been introduced, no previous work focuses on how to compute gain information on parallel architectures.  In this thesis, we will focus on how to do this and what improvement we can get.