We investigated the performance of the handheld radial shape discrimination (hRSD) test in detecting the development of neovascular AMD (nAMD) in a prospective, longitudinal, observational study. Patients diagnosed with unilateral nAMD, with no nAMD in the other eye (the study eye, SE), completed the hRSD test on consecutive, routine clinic visits up to a maximum of 12, or until they were diagnosed with nAMD in the SE based on slit-lamp biomicroscopy and spectral-domain OCT assessment, with fluorescein angiography confirmation. Masked grading was carried out to confirm the diagnosis of nAMD, and to ensure no cases of nAMD were missed. Receiver operating characteristics (ROC) analysis was used to explore the diagnostic performance of the hRSD test relative to clinical diagnosis. Data were available from 179 patients of whom 19 (10.6%; "converters") developed nAMD in the SE. The mean hRSD threshold at conversion was -0.47 (95% CI -0.38 to -0.55) logMAR compared to -0.53 (-0.50 to -0.57) logMAR in 160 non-converters. hRSD threshold in the converters began to decline 190 days before diagnosis of nAMD. The ROC curve demonstrated that at an hRSD cut-off of -0.60 logMAR, sensitivity was 0.79 (0.54-0.94) with a specificity of 0.54 (0.46-0.62); positive and negative predictive values were 0.16 and 0.96 respectively. We conclude that the hRSD test has moderate sensitivity for detecting the earliest stages of nAMD in the at-risk fellow eyes of patients with unilateral nAMD, compared to clinical diagnosis. Given its relative inexpensiveness, ease of use and the inherent connectivity of the platforms it can be presented on, it may have a role in early detection of nAMD in the population at large.