The microtubule-isolation procedures described here are based on the ability of the investigator to control the dimer-polymer distribution of tubulin by varying the temperature of the extract. In general, the extract is warmed to induce microtubule assembly, the polymer is collected by centrifugation, cooled to induce disassembly, clarified by centrifugation, and then warmed again to produce polymer. As long as the GTP supply is sufficient, the microtubules that result can be taken through numerous rounds of this in vitro assembly and disassembly reaction. Many microtubule-associated proteins (MAPs) associate with microtubules assembled in vitro. Some reagents can skew the equilibrium of assembly and disassembly toward formation of polymer. The inclusion of glycerol, for instance, promotes microtubule assembly by disrupting the hydration shell around the tubulin dimers. The result is a greater yield of tubulin per gram of starting material. However, the ratio of MAPs to tubulin is slightly lower, presumably because the glycerol also decreases the binding of MAPs to tubulin. Two methods are described here: The first uses buffer lacking assembly-promoting components, and the second uses buffer containing glycerol. These procedures are most efficient with vertebrate brain tissue, where the soluble protein can be up to 15%-20% tubulin. The first produces satisfactory yields when using chick or pig brain; the second is recommended for calf or cow brain. The second procedure may also be useful for studies of nonneuronal tissues where the relative concentration of tubulin per gram of wet weight is considerably lower than that of brain.