Participants judged the direction of the frequency difference between a standard tone and a comparison tone separated by a silent interval and preceded by a series of three interfering tones. The frequencies of the interfering tones were all either lower/higher than (providing interference) or the same as (providing no interference) the standard-tone frequency. When providing interference, the interfering tones were further ordered either randomly or so that they formed melodically ascending/ descending sequences toward the standard-tone frequency. Irrespective of the order of the interfering tones, the judgments were more accurate when the interfering tones and the comparison tone deviated in frequency in the opposite, rather than the same, direction from the standard tone. Reaction times of these judgments were in line with the data based on the judgment accuracy. The results suggest that the representations of individual interfering tones, and not of anticipations extrapolated from them as a compound, were involved in proactive interference with frequency-comparison performance.