Traditional decision tree classifiers work with the data whose values are known and precise. We can also extend those classifiers to handle data with uncertain information. Value uncertainty arises in many applications during the data collection process. Example sources of uncertainty measurement/quantization errors, data staleness, and multiple repeated measurements. Rather than abstracting uncertain data by statistical derivatives, such as mean and median, the accuracy of a decision tree classifier can be improved much if the complete information of a data item is used by utilizing the Probability Density Function (PDF). In particular, an attribute value can be modelled as a range of possible values, associated with a PDF. The PDF function has only addressed simple queries such as range and nearestneighbour queries. Queries that join multiple relations have not been addressed with PDF. Despite the significance of joins in databases, we address join queries over uncertain data. We propose semantics for the join operation, define probabilistic operators over uncertain data, and propose join algorithms that provide efficient execution of probabilistic joins especially threshold. In which we avoid the semantic complexities that deals with uncertain data. For this class of joins we develop three sets of optimization techniques: item-level, page-level, and index-level pruning. We will compare the performance of these techniques experimentally.