According to a transfer-appropriate processing framework, immediate priming costs arise from a match between a prime and probe event on 1 dimension and a difference between those 2 events on some other dimension (i.e., a partial match). In Experiment 1, the authors used a Stroop priming procedure to generate 6 variants of partial match, yet only 1 of these 6 conditions yielded a partial match cost. Experiment 2 demonstrates that, for some of these conditions, an underlying partial match cost was obscured by the contribution of an independent source of facilitation to performance. In Experiment 3, however, a partial match was observed to have produced an immediate priming cost only when the selective attention demands of the probe task were high. Overall, the results reveal a limitation in current applications of the transfer-appropriate processing framework to immediate priming phenomena. (Contains 3 footnotes, 4 tables, and 3 figures.)