Axons and dendrites are distinguished by microtubule polarity. In Drosophila, dendrites are dominated by minus-end-out microtubules, whereas axons contain plus-end-out microtubules. Local nucleation in dendrites generates microtubules in both orientations. To understand why dendritic nucleation does not disrupt polarity, we used live imaging to analyze the fate of microtubules generated at branch points. We found that they had different rates of success exiting the branch based on orientation: correctly oriented minus-end-out microtubules succeeded in leaving about twice as often as incorrectly oriented microtubules. Increased success relied on other microtubules in a parallel orientation. From a candidate screen, we identified Trim9 and kinesin-5 (Klp61F) as machinery that promoted growth of new microtubules. In S2 cells, Eb1 recruited Trim9 to microtubules. Klp61F promoted microtubule growth in vitro and in vivo, and could recruit Trim9 in S2 cells. In summary, the data argue that Trim9 and kinesin-5 act together at microtubule plus ends to help polymerizing microtubules parallel to pre-existing ones resist catastrophe.