Heart disease commonly occurring disease and is the major cause of sudden death nowadays. This disease attacks the persons instantly. Most of the people do not aware of the symptoms of heart disease. Timely attention and proper diagnosis of heart disease will reduce the mortality rate. Medical data mining is to explore hidden pattern from the data sets. Supervised algorithms are used for the early prediction of heart disease. Nearest Neighbor (KNN) is the widely used lazy classification algorithm. KNN is the most popular, effective and efficient algorithm used for pattern recognition. Medical data sets contain 14 features is obtained from UCI Machine Learning Repository. Feature subset selection is proposed to solve this problem. Feature selection will improve accuracy and reduces the running time. This paper investigates to apply KNN for prediction of heart disease. Experimental results show that the algorithm performs very well with 86% accuracy. This system also provides the relation between diabetes and how much it influences heart disease