Thirty smokers and 30 nonsmokers participated in a flicker study in which the role of attentional bias in change detection was examined. The participants observed picture pairs of everyday objects flicker on a computer screen until they detected the one object that had changed. In half of the pictures, a smoking-related object (e.g., a lighter) was included among smoking-unrelated objects (e.g., a spoon). Half of the smokers and half of the nonsmokers were aware of the experiment's focus, and the other half were not. The smokers exhibited shorter detection latencies than did the nonsmokers when a smoking object changed and longer detection latencies when a nonsmoking object changed while a smoking object was present, and they exhibited detection latencies similar to those of the nonsmokers when smoking objects were not present. Interestingly, the nonsmokers displayed the same attentional bias as the smokers when they were aware of the experiment's smoking focus, but they did not display any attentional bias when they were unaware. Thus, these findings provide evidence for long-term context-independent, as well as for short-term context-dependent, attentional bias.