Existing models of ambivalence suggest that as the number of conflicting reactions (e.g., attitude components) increases, so too does the experience of ambivalence. Interestingly, though, these models overwhelmingly assume that this relationship is independent of valence. Across three studies we observe that this effect is in fact heavily influenced by two established valence asymmetries: positivity offset (baseline positive reactions even in the absence of positive information) and negativity bias (greater impact of negative reactions than positive reactions). Consistent with positivity offset, we observe that subjective ambivalence is greater when people have univalent negative rather than univalent positive attitudes. However, as conflicting information is acquired, subjective ambivalence rises more quickly when that information is negative rather than positive. The latter effect is consistent with negativity bias and suggests that although people feel more conflicted when they have only negative (versus only positive) reactions, they also feel more conflicted when they have mostly positive (versus mostly negative) reactions. Our investigation also uncovers an interesting consequence of these asymmetries: When people have mixed reactions, they do not experience maximum ambivalence at equal levels of positivity and negativity, as suggested by canonical ambivalence theory. Rather, subjective ambivalence peaks when positive reactions outnumber negative reactions. These effects are found to have downstream consequences for other dimensions of attitude strength.