The ability of actin filaments to function in cell morphogenesis and motility is coupled to their capacity for rapid assembly and disassembly. Because disassembly in vitro is much slower than in vivo, cellular factors that stimulate disassembly have long been assumed to exist. Although numerous proteins can affect actin dynamics in vitro, demonstration of in vivo relevance of these effects has not been achieved. We have used genetics and an actin-inhibitor in yeast to demonstrate that rapid cycles of actin assembly and disassembly depend on the small actin-binding protein cofilin, and that cofilin stimulates filament disassembly. These results may explain why cofilin is ubiquitous in eukaryotes and is essential for viability in every organism in which its function has been tested genetically. Magnitudes of disassembly defects in cofilin mutants in vivo were found to be correlated closely with the magnitudes of disassembly defects observed in vitro, supporting our conclusions. Furthermore, these cofilin mutants provided an opportunity to distinguish in living cells those actin functions that depend specifically on filament turnover (endocytosis) from those that do not (cortical actin patch motility).