The present research investigated whether the affective priming paradigm from Fazio, Sanbonmatsu, Powell, and Kardes (1986) can be used as an implicit measure of person schemata. Names and faces of friends or romantic partners and of a disliked person were used as primes. It was explored whether: (1) stimuli relating to liked and disliked persons elicit congruency priming effects similar to those reported for words; (2) masked and unmasked priming procedures had similar effects; and (3) whether individual differences in the implicit measure were related to explicit measures of relationship quality. For clearly visible primes the expected congruence priming effects were found across names and faces. For marginally visible primes, however, unexpected reverse priming effects were observed for the disliked person. In a second experiment, a confound of the familiarity and evaluation of the significant other primes was removed. Now a reverse priming effect could be demonstrated for masked primes in both liked a...