The processing of chords is facilitated when they are harmonically related to the context in which they appear. The purpose of this study was to assess whether this harmonic priming effect depends on the version (normal vs. scrambled) of the context chord sequences. Normal sequences were scrambled by permuting chords two-by-two (Experiment 1) or four-by-four (Experiments 2 and 3). Normal chord sequences were judged less coherent than scrambled sequences. However, normal chord sequences showed facilitation for harmonically related rather than for unrelated targets, and this effect of relatedness did not diminish for scrambled sequences (Experiments 1-3). The data of musicians and nonmusicians were interpreted with Bharucha's (1987) spreading activation framework. Simulations suggested that harmonic priming results from activation that spreads via schematic knowledge of Western harmony and accumulates in short-term memory over the course of the chord sequence.