The normal aged group who had never suffered from depression, and the aged depressed group were recruited using GDS, SDS and HAMD as tools. The cue-target task and emotional Stroop task were used to test the phases of attention bias in depression, and the effect of depth of processing and the cues on the attention biases in depressive older people. The results showed that: (1) In the cue-target task, when the presentation of stimuli was for 250ms or 500ms, there were no differences in IOR effects between all types of stimulation for both depressed aged people and the normal aged group (P> 0.05) under this condition, this means that the aged depressed group showed no attention bias. When the presentation of stimuli was for 750ms or 1000ms, the depressed aged people's IOR effects were significantly different under different types of stimulation, F(2, 66) = 3.22, P = 0.046, while the normal group showed no difference in IOR effects under all types of stimulation (P> 0.05) . Further test indicated that when the presentation of stimuli was for 750ms or 1000ms, the IOR effects to the stimulation of sad faces were significantly different between the two groups, F(1, 33) = 10.03, P = 0.003; F(1, 33) = 18.21, P = 0.000. These indicated that the depressed aged people's attentional biases appeared when the presentation of stimuli was 750ms or 1000ms. (2) In the Stroop task, there were significant differences of RTs under different cue types of stimulation in the aged depression group and there were significant differences of RTs between two groups under the stimulation of sad faces, double sad faces, sad and natural faces, F(1, 41) = 5.37，P=0.026；F(1, 41) = 7.08，P=0.011；F(1, 41) = 7.28，P=0.010；which hints that attention bias in depressive aged group to negative cues may appear in early attentional process phase. The authors considered that the depressed aged people’s attention bias can occur in the initial allocation phase or attention disengagement phase; and was impacted by the depth of processing.