Why are B-trees used in databases instead of binary trees?

B-trees are optimized for disk-based storage where I/O is the bottleneck. They have large branching factor (hundreds of children per node) matching disk block size, minimizing tree height and disk reads. A B-tree with 1000-way branching and millions of keys has height 2-3 (only 2-3 disk reads). They maintain balance automatically, support range queries efficiently, and keep data sorted. B+ trees, a variant storing data only in leaves, are most common in databases.