Electoral campaigns and policy debates occur over-time. Throughout the campaign, competing sides push opposing frames and which frame proves most effective often determines who wins. Success hinges, in large part, on launching influential counter-frames – that is, frames that oppose earlier effective frames. Surprisingly, scholars have paid virtually no attention to the dynamics of counter-framing. In this paper, we explore how timing and repetition of counter-frames affect their success. Our results – based on an over-time experiment – show that the best counter-framing strategy is highly contingent on the nature of the respondents. For those who form initially strong opinions, the counter-frames are most effective when delayed in time, and perhaps surprisingly, not repeated. In contrast, the lag time between the initial frame and a counter-frame is irrelevant for those who form weak opinions, but repetition is often beneficial. Our findings highlight the contingent nature of successful framing and isolate sources of political power in campaigns.