The current study investigated the antecedents and consequences of peer-rated intelligence in a longitudinal round robin design, following previously unacquainted members of small student work groups. Results indicated that peer-reputations of intelligence were reliable, stable and weakly correlated with objective intelligence. Bias was shown by correlations with interpersonal liking (decreasing across time) and idiosyncratic rating tendencies (increasing across time). Agreement between self-ratings and peer-reputations increased over time but was not based on increasing accuracy but on reciprocal associations between self-ratings and peer-reputations in the beginning of the acquaintanceship process, and on peer-reputations predicting changes in self-ratings later on. Finally, it was shown that peer-rated intelligence reputations predict academic achievement across two 4-month periods (even when tested intelligence was controlled) and dropout from university after 8 months. Overall, the pattern of results demonstrates the utility of a socioanalytic perspective in analysing personality and social processes. Copyright © 2010 John Wiley & Sons, Ltd.