Reasoning about bedrock abstract concepts such as time, number, and valence relies on spatial metaphor and often on multiple spatial metaphors for a single concept. Previous research has documented, for instance, both future-in-front and future-to-right metaphors for time in English speakers. It is often assumed that these metaphors, which appear to have distinct experiential bases, remain distinct in online temporal reasoning. In two studies we demonstrate that, contra this assumption, people systematically combine these metaphors. Evidence for this combination was found in both directly elicited (Study 1) and spontaneous co-speech (Study 2) gestures about time. These results provide first support for the hypothesis that the metaphorical representation of time, and perhaps other abstract domains as well, involves the continuous co-activation of multiple metaphors rather than the selection of only one.