The molecular basis of microtubule dynamic instability is controversial, but is thought to be related to a "GTP cap." A key prediction of the GTP cap model is that the proposed labile GDP-tubulin core will rapidly dissociate if the GTP-tubulin cap is lost. We have tested this prediction by using a UV microbeam to cut the ends from elongating microtubules. Phosphocellulose-purified tubulin was assembled onto the plus and minus ends of sea urchin flagellar axoneme fragments at 21-22 degrees C. The assembly dynamics of individual microtubules were recorded in real time using video microscopy. When the tip of an elongating plus end microtubule was cut off, the severed plus end microtubule always rapidly shortened back to the axoneme at the normal plus end rate. However, when the distal tip of an elongating minus end microtubule was cut off, no rapid shortening occurred. Instead, the severed minus end resumed elongation at the normal minus end rate. Our results show that some form of "stabilizing cap," possibly a GTP cap, governs the transition (catastrophe) from elongation to rapid shortening at the plus end. At the minus end, a simple GTP cap is not sufficient to explain the observed behavior unless UV induces immediate recapping of minus, but not plus, ends. Another possibility is that a second step, perhaps a structural transformation, is required in addition to GTP cap loss for rapid shortening to occur. This transformation would be favored at plus, but not minus ends, to account for the asymmetric behavior of the ends.