Evidence regarding the ability of attention to bias neural processing at the level of single features has been gathering steadily, but most of the experiments to date used arrays with multiple objects and locations, making it difficult to rule out indirect influences from object or spatial attention. To investigate feature-specific selective attention, we have assessed the ability to select and ignore individual features within the same object. We used a negative-priming paradigm in which the color or the direction of internal motion of the object could determine the relevant response. Bidimensional (colored and moving) and unidimensional (colored and stationary, or gray and moving) stimuli appeared in unpredictable order. In successive blocks, participants were instructed that one feature dimension was dominant. During that block, participants responded according to the dominant dimension for bidimensional stimuli. For unidimensional stimuli, participants responded to the only dimension of the stimulus that afforded a response, regardless of the instruction for the block. The ability to inhibit irrelevant task information at the level of specific features (negative priming for features) was indexed by a decrease in performance to detect one particular feature value (e.g., red) if the same feature value (red) but not another color value (green) had been ignored in the previous bidimensional stimulus. Behavioral results confirmed the existence of inhibitory, negative-priming mechanisms at the single-feature level for both color and motion dimensions of stimuli. Event-related potentials recorded during task performance revealed the dynamics of neural modulation by feature attention. Comparisons were made using the identical physical stimuli under different conditions of attention to isolate purely attentional effects. Processing of identical bidimensional stimuli was compared as a function of the dimension of attention (color, motion). Processing of identical unidimensional stimuli that followed bidimensional stimuli was also compared to identify possible effects of feature-specific negative priming. The electrophysiological effects revealed that inhibition of irrelevant features leads to modulation of brain activity during early stages of perceptual analysis.