A randomized block design is an experimental design in which experimental units are first grouped into blocks that are similar with respect to a relevant characteristic and treatments are then randomly assigned within each block. Blocking removes block-to-block variation from the treatment comparison. A randomized complete block design is the special case in which every treatment occurs in every block.
This statistical use of "block" differs from a block design, where each block is a subset of treatments or points. The viewpoints meet in an incomplete block design, in which each experimental block receives only a subset of the treatments.
