High noise tolerance is a hallmark of sensory systems as external stimuli are intrinsically noisy. However, the neural mechanisms underlying the noise tolerance are not clearly understood. In Drosophila vision, we evaluated behavioral and physiological changes in visual responses to simple visual patterns––translating bars, gratings, spots, and looming objects––overlaid with varying levels of salt-and-pepper noise. In tethered, flying Drosophila, we found that the noise tolerance, evaluated by changes in visually evoked wingbeat responses, depends on the pattern. Namely, the noise tolerance was highest for the grating patterns, medium for the bar and the looming disc patterns, and lowest for the spot patterns. Furthermore, we found that wing response latencies increased significantly for increasing noise levels for the spot and loom patterns, but not for the grating and bar patterns, suggesting distinct neural pathways associated with these visual patterns. To evaluate the noise tolerance at the neuronal level, we genetically silenced a group of visual neurons known to be associated with these visual patterns as well as two neuromodulatory neurons. We found the largest reduction in wing responses to the loom pattern when either a specific group of visual projection neurons (LPLC2) or octopaminergic neurons were silenced. To a noiseless loom pattern, however, silencing of either LPLC2 or octopaminergic neurons had little, if any, effect on the wing responses, suggesting the role of these neurons in noise suppression. Finally, we measured physiological responses of the visual neurons via calcium imaging and found that the noise tolerance of individual cell types was consistently lower than that of behavioral responses, suggesting further noise suppression in the downstream circuits. This work will lead to the understanding of the neural mechanisms by which sensory systems extract salient stimulus features in a noisy environment.