Binding accounts propose that whenever we want to execute an action, we must first bind all task-relevant percepts and motor programs together to derive a working action plan. These bindings outlast action execution and can influence subsequent actions, leading to so-called binding effects. Thus, if preexisting bindings and a newly generated action plan coincide, action performance benefits are observed. However, if preexisting bindings and a newly generated action plan do not coincide, because features only partially overlap, performance costs are observed. So far, laboratory studies have focused on binding effects resulting from a single preceding action episode. However, natural behavior is almost always composed of a string of action episodes, all of which in theory generate bindings. Because bindings are stored in limited working memory space, multiple active bindings should interfere with one another. However, in what manner, is so far unclear. Concretely, binding competition might depend on the quality of the bound elements, or it might be independent of the quality of the bound items. We employed a response-response binding paradigm with a prime-intervening-trial-probe design to investigate residual binding effects in a probe, after an additional intervening action pair that was either a full change or a partial repetition from the prime. While we found that intervening-trials in general reduced binding effects in the probe, the type of intervening-trial did not substantially influence probe binding effects. This is clear evidence for content independent interference between bindings that most likely results from an overall limit in working memory capacity.