A curve is said to be **self-similar** if, for every piece of the curve, there is a smaller piece that is similar to it. For instance, a side of the Koch snowflake is self-similar; it can be divided into two halves, each of which is similar to the whole.

An interesting result in telecommunications traffic engineering is that packet switched data traffic patterns seem to be statistically self-similar.

See also:

- fractal
- scale invariance