A block is a group of experimental units that share some characteristic. Thus blocks will differ, but experimental material within a block will be relatively more uniform. The blocking concept is the primary method used to control for known sources of variation in an experiment. Often fields or years are blocks, because experimental material within a field or year is more uniform, and by blocking the differences among fields or years can be controlled.

Of course, to block on something, the researcher must be able to identify sources of variability before the experiment is conducted.

Blocks can be either complete or incomplete. Complete blocks have enough experimental units to test all the treatments. Incomplete blocks are not large enough to hold all the treatments, which can produce an incomplete block design.

Covariates are another approach to controlling for variability.