Examined the effects of motivated processing on hindsight bias. One hundred fifty three college students estimated and later recalled the likelihood that 30 self-relevant events would occur during the next 2 months. Multi-level modelling was used to determine the (within-subject) effects of expectations, event valence and event controllability on hindsight bias and the extent to which these effects were moderated by participants' need for cognition (NFC) scores (between-subjects). For events that actually occurred, we found support for defensive processing in that the bias was smaller for negative events. Also, for events that actually occurred, those judged as more controllable produced a larger bias. Neither valence nor controllability had any effect on the size of the bias for events that did not occur. The size of the bias for occurrences did not differ significantly from that for non-occurrences. Finally, NFC did not moderate the effects of valence or controllability, nor did it directly affect the size of the bias.