We present a new corpus of 147 high-quality photographic colour images (the Hatfield Image Test: HIT). Existing sets of pictorial stimuli tend to be line drawn, contain many items that are readily identifiable by healthy participants, and, therefore, have an inherent tendency towards ceiling effects in the normal population. The broad range of item difficulty and range of semantic subcategories in the HIT permits researchers to select stimuli of appropriate difficulty as required. We present naming data from 152 healthy participants. Additionally, we present mean ratings for each item on several widely used psycholinguistic variables: age of acquisition, colour diagnosticity, familiarity, name agreement (and the H statistic), visual complexity, and word frequency. These stimuli provide a useful corpus for experimental and clinical researchers.