This paper investigates an implication of the self-serving bias for reciprocalresponses. It is hypothesized that negative intentionality matters more thanpositive intentionality for reciprocating individuals with a self-servingattributional style. Experimental evidence obtained in the hot response gamesupports this prediction. Subjects are 67% more likely to reciprocate anintentional hurtful choice over an unintentional hurtful choice. Subjects areonly 25% more likely to reciprocate an intentional helpful choice over anunintentional helpful choice. The evidence on the intermediating role ofemotions is consistent with the explanation offered by the self-serving bias.