This paper reports three experiments which examined individual differences in the semantic domain of bird names. In Experiment 1 each subject emitted his own list of 20 bird names and then scaled them using a rating scale version of the repertory grid test. This test requires that the subject generate a set of bipolar descriptive scales for the purposes of the scaling. Principal components analysis of the resulting data matrices were used to construct a semantic space for each of the subjects. Interpretation of the first three components for each subject suggested that most discriminate water and land birds; predatoriness was also a commonly found dimension. In Experiment 2 each subject classified each of the bird names s/he had listed in terms of the bipolar scales which arose during the scaling. As predicted, judgment reaction time (RT) was a U-shaped function of the location of the bird name on the bipolar scale. In Experiment 3 categorization of bird names was faster when a semantically close bird name was used as a prime than when a semantically distant bird name was the prime. These results demonstrate the practicability of an individual differences approach to the study of semantic memory and show that individual semantic spaces can be used to predict RT performance in semantic memory tasks.