This thesis focuses on extracting relationships from patient narratives. The Clinical E-Science Framework (CLEF) project was used to extract important information from medical texts by building a system for the purpose of clinical research, evidence-based healthcare and genotype-meets-phenotype informatics. The system is divided into two parts, one part concerns with the identification of relationships between clinically important entities in the text. The full parses and domain-specific grammars had been used to apply many approaches to extract the relationship. In the second part of the system, statistical machine learning (ML) approaches are applied to extract relationship. A corpus of oncology narratives that hand annotated with clinical relationships can be used to train and test a system that has been designed and implemented by supervised machine learning (ML) approaches. Many features can be extracted from these texts that are used to build a model by the classifier. The goal of this thesis is to analyze the implementation of multiple supervised learning algorithms for relationship extraction. As well as the effects of adding the features, changing the size of the corpus, and changing the uneven margin parameter of the algorithm on relationship extraction are examined. Also the evaluation of the supervised learning algorithms and comparing them based on the performance. The results of the performance of these algorithms for extracting clinical relationships from medical text are very important. SVM with uneven margin is much suitable algorithm which achieves high accuracy, but it takes more time in the run than Perceptron with uneven margin. Perceptron with uneven margin is very fast algorithm than others as well as the accuracy is relatively near to SVM, there is small change in between. Increasing the value of τ leads to improve the performance to reach the value that has high performance where τ = 0.8 after that point the performance dropped. Adding new feature sets like non-syntactic features improves the performance. Adding the syntactic features leads to small drop in the performance that unclear. Changing the size of training corpus leads to improve the performance although required increasing the classification time.