Previous studies have shown that attentional bias modification (ABM) is effective in reducing negative attentional biases. However, the mechanisms underlying how ABM effectively reduces negative attentional biases are still unclear. In the present study, we conducted an ABM procedure that included a 3-day training session with a sample of nonclinical participants (N = 40) to investigate the effect of ABM on emotional and nonemotional attentional biases. Participants completed a modified dot-probe task with 2 different instructions (explicit or standard) during the training; their attentional biases were tested before and after the training. Only participants trained with explicit instructions showed a reduction in negative attentional biases in dot-probe task and an improvement in attentional disengagement from negative stimuli in gap-overlap task. On the other hand, attention toward nonemotional stimuli was only marginally improved by training with both explicit and standard instructions. These results indicate that explicit instructions may promote ABM training.