A fundamental problem in cell biology is how cells control the size of their organelles. The limiting pool mechanism, where assembly is controlled by the depletion of a limited pool of components, successfully explains experiments on nucleoli, centrosomes, and spindles. This mechanism however cannot explain how cells assemble filaments whose length is matched to the length of the cell. In this talk I will describe our experiments on actin cables in yeast cells and related theory that address the question of how cells control the assembly of a filament to a specified length. A simple stochastic model of cable assembly leads to a Gumbel distribution of cable lengths consistent with our experiments. Published data on length fluctuations of filamentous structures in a variety of cells shows universal features predicted by extreme value statistics. These findings point to a more general theory of control of assembly in cells that is geometrical in nature and independent of specific biochemical reactions.