Repetition priming of visual search traditionally refers to the reductions in reaction time when a target‚ or a target feature‚ is repeated on consecutive trials. Priming was initially thought to be a facilitated response to the target-defining feature, while repetition of irrelevant target features was thought not to benefit visual search. The simple feature facilitation view has later been contested by results where repetition priming was only observed in response to whole-target repetition and results where search-irrelevant features were primed independently of the target-defining feature. Recent proposals assume that repetition priming operates at multiple levels. Here we focus on two questions: (1) does repetition priming have a purely perceptual component? and (2) how does such a component fit into the TVA framework? (Bundesen, C. 1990. Psychological bulletin and Review, 97, 523-547) In this study we briefly (20-110 ms.) presented subjects with a circular array of 6 digits, one of which was a color singleton target (1T5D partial report), whole report trials (6T) where subjects reported as many uniformly colored digits as possible and single target presentation (1T0D). Non-speeded accuracy tasks were chosen to isolate the perceptual component in repetition priming from response related biases and motor components. Color and position repetition benefits were evident at very brief exposures. The results support the hypothesis that repetition priming occurs for feature processing, while not ruling out repetition benefits at other levels. Furthermore, color repetition was only beneficial when the task required attentional selection. The results were fitted to a TVA-based model from which we provided estimates of weights. The results from the modeling procedure indicate that the perceptual component of repetition priming is best described as a modification of feature weightings by recently attended features.