At metaphase, the amount of tubulin assembled into spindle microtubules is relatively constant; the rate of tubulin association equals the rate of dissociation. To measure the intrinsic rate of dissociation, we microinjected high concentrations of colchicine, or its derivative colcemid, into sea urchin embryos at metaphase to bind the free tubulin, thereby rapidly blocking polymerization. The rate of microtubule disassembly was measured from a calibrated video signal by the change in birefringent retardation (BR). After an initial delay after injection of colchicine or colcemid at final intracellular concentrations of 0.1-3.0 mM, BR decreased rapidly and simultaneously throughout the central spindle and aster. Measured BR in the central half-spindle decreased exponentially to 10% of its initial value within a characteristic period of ~20 s; the rate constant, k = 0.11 ± 0.023 s-1, and the corresponding half-time, t( 1/2 ), of BR decay was ~6.5 ± 1.1 s in this concentration range. Below 0.1 mM colchicine or colcemid, the rate at which BR decreased was concentration dependent. Electron micrographs showed that the rapid decrease in BR corresponded to the disappearance of nonkinetochore microtubules; kinetochore fiber microtubules were differentially stable. As a control, lumicolchicine, which does not bind to tubulin with high affinity, was shown to have no effect on spindle BR at intracellular concentrations of 0.5 mM. If colchicine and colcemid block only polymerization, then the initial rate of tubulin dissociation from nonkinetochore spindle microtubules is in the range of 180-992 dimers per second. This range of rates is based on k = 11% of the initial polymer per second and an estimate from electron micrographs that the average length of a half-spindle microtubule is 1-5.5 μm. Much slower rates of tubulin association are predicted from the characteristics of end-dependent microtubule assembly measured previously in vitro when the association rate constant is corrected for the lower rate of tubulin diffusion in the embryo cytoplasm. Various possibilities for this discrepancy are discussed.