Purpose. Histopathologic and morphometric studies on donor retinas with AMD indicate a predilection for loss of rods over cones in the early stages of disease 3. We examined whether a psychophysical correlate of selective rod vulnerability could be demonstrated. Methods. Subjects were 53 older adults (59-87 yrs, VA 20/14-87) with non-exudative AMD in the tested eye and 11 older adults (66-86 yrs, VA 20/17-33) in normal retinal health. Assignment to AMD vs. normal groups was based on grading of fundus photographs. Using a modified Humphrey Field Analyzer, photopic sensitivity was assessed with a 600nm stimulus on a 10 cd/m 2 background; scotopic sensitivity was measured with a 500nm stimulus following 40 min of dark adaptation. The 1.7° stimulus was presented at 52 locations throughout the central field (30° radius). All subjects were dilated prior to testing. Thresholds were individually corrected for lens density. Data from the old-normal group were used to define sensitivity loss in the AMD group. Results. For 79% of AMD patients, average scotopic sensitivity loss across the field was greater than photopic sensitivity loss (mean loss 1 IdB vs. 3dB respectively); for 11%, photopic sensitivity loss was worse, and for 10%, they were equal. For 68% of AMD patients, scotopic sensitivity values at 2 or more contiguous test points along the horizontal meridian fell > 5dB below the normal range (95% CI for old-normals). Those AMD patients with larger average scotopic impairments reported being more bothered by night vision problems in a symptom survey and having more difficulty on the night-driving subscale of the Activities of Daily Vision Scale (p<.0005). Conclusions. Rod-mediated sensitivity loss in AMD tends to be more severe than is cone-mediated sensitivity loss, thus providing a psychophysical correlate to anatomic findings that rods are selectively vulnerable in the earlier stages of this condition. Furthermore, patients' symptoms and difficulties in daily visual activities under low illumination reflect this vulnerability.