This study seeks to advance the literatures on organizational improvisation and unexpected events. It tackles the question of whether the relative presence of improvisation during a startup's response to an ordinary, unexpected event affects the value of that response, an issue of clear importance given the ubiquity of unexpected events in startups. Improvisation in practice typically involves varying degrees of predesigned and extemporaneously designed activity. The study explores the dangers of simultaneously mixing predesigned actions and improvisational activity. It develops theory in the context of startups' action streams in response to 141 unexpected events identified by field informants. Results from hypothesis tests support theory that the relative presence of improvisation in an action stream in response to an unexpected event will have a U-shaped impact on its success resolving that event: a mixed presence shows relatively poorer outcomes than either concentrated predesigned action or a high presence of improvisation. The study also extends prior work by theorizing and finding evidence that two sources of organizational memory-firm-specific experience (proxied by organizational age) and nonfirm-specific experience (proxied by founders' business experience prior to founding)-moderate the value of the presence of improvisation in response to unexpected events in different ways, consistent with greater challenges to rapidly integrating varied knowledge. Finally, it contributes to understanding of improvisation patterns in response to ordinary, unexpected events, suggests areas for additional research, and offers managerial implications for startups such as the value of deliberately raising shared awareness of shifts to organizational improvisation.