Medicine is a substance that is used to diagnose, eliminate, and cure diseases, injuries, or others in humans. Handling and prevention of various diseases cannot be separated from therapeutic actions with drugs. Drug grouping serves to classify drugs into several groups to determine the characteristics of a drug or not. By knowing the characteristics of each existing drug, it can be easier to determine an effective marketing pattern. The use of data mining can help to cluster drugs by utilizing existing sales data. In this study the methodology used is CRISP-DM with the stages carried out namely Business Understanding, Data Understanding, Data Preparation, Modeling, Evaluation, and Deployment. The dataset used is Amarta Sehat Pharmacy data from January-December 2021. The K-Means algorithm is used for cluster formation using Jupyter Notebook tools with the python programming language. The elbow method serves to determine the best number of clusters (K), the recommendation from the elbow method produces the 5 most optimal clusters and is also calculated by evaluating the Sum of square error with an optimal cluster value of 7154215036292.542. The results of drug Clustering obtained to determine an effective marketing pattern at the Amarta Sehat Pharmacy are 11 drugs that are classified as high-selling drugs, 76 drugs are classified as best-selling drugs, 131 drugs are classified as drugs with the category of selling well. quite in demand, 399 drugs into the category of drugs that are not in demand, and 326 drugs into the category of drugs that are not in demand