Perceiving and reacting to multidimensional objects creates so-called event files via feature binding. Bogon, Thomaschke, and Dreisbach (2017) provided first evidence for the integration of the feature stimulus duration into such event-files. However, their paradigm did not allow to differentiate between stimulus-stimulus and stimulus-response binding. The present study used a classification task with many-to-one mappings that allows to examine the integration of stimulus and response features independently. Experiment 1 used a pitch classification task. Participants had to respond with a left keypress to a low and a very low sine-tone and with a right keypress to a high and very high sine-tone. The four sine-tones were presented for either a short or long duration, resulting in a total of eight stimuli. As an indicator of binding, we used partial repetition costs (better performance when both pitch/response and duration repeat or shift relative to partial repetitions). Results of Experiment 1 indicate stimulus-response binding and no stimulus-stimulus binding. In Experiment 2, instead of classifying the pitch of artificial sine-tones, participants had to classify the type of musical instruments that produced the stimulus tones. Results replicated evidence for stimulus-response binding, but also provided evidence for stimulus-stimulus binding. The potential relevance of duration in a musical context is discussed as a potential moderator of stimulus-stimulus binding.