Brain tumor is a disease that emerge due to the growth of abnormal cells in the brain, which can interfere with the brain function. There are two types of brain tumors, first, primary brain tumors, where the abnormal cells grow in the brain, second, secondary brain tumors, where the abnormal cells growth originates from other organs that spread to the brain. Early-stage brain tumors are often not detected, so when they are known to exist, they have entered a high stage. Therefore, the importance of prevention or early detection of brain tumors is crucial. One way of prevention is the implementation of machine learning. This study presents a comparison of the accuracy of K-Nearest Neighbor (KNN) and Extreme Learning Machine (ELM) in the diagnosis of brain tumors. K-Nearest Neigboris a supervised learning algorithm that classifies new instances based on the majority of k-neighbors. Extreme Learning Machine is a feedforward neural network algorithm that has one hidden layer to determine the input and output load values. The results showed that the extreme learning machine was superior in diagnosing brain tumors with an accuracy rate of 97.3% compared to k-nearest neighbor with an accuracy rate of 81.5%./Tumor otak merupakan penyakit yang timbul akibat tumbuhnya sel-sel abnormal pada otak, yang dapat menggangu fungsi kerja otak. Tumor otak memiliki dua jenis, yaitu tumor otak primer, dimana pertumbuhan sel abnormal di otak dan tumor otak sekunder, dimana pertumbuhan sel abnormal berasal dari organ lain yang menyebar ke otak. Tumor otak stadium awal, seringkali tidak terdeteksi keberadaan, sehingga ketika sudah diketahui keberadaannya, sudah memasuki stadium tinggi. Oleh karena itu, pentingnya pencegahan atau deteksi awal terhadap tumor otak. Salah satu cara pencegahan adalah implementasi machine learning. Penelitian ini menyajikan perbandingan tingkat akurasi K-Nearest Neighbor (KNN) dan Extreme Learning Machine (ELM) dalam diagnosis penyakit tumor otak. K-Nearest Neighbor merupakan algoritma supervised learning yang mengklasifikasikan instance baru berdasarkan mayoritas dari k-tetangga. Extreme Learning Machine adalah algoritma feedforward neural network yang memiliki 1 lapisan tersembunyi untuk menentukan nilai beban input dan output. Hasil penelitian menunjukkan bahwa Extreme Learning Machine lebih unggul dalam diagnosis penyakit tumor otak dengan tingkat akurasi 97.3% dibandingkan K-Nearest Neighbor dengan tingkat akurasi 81.5%.