PURPOSE: To assess whether baseline retinal nerve fiber layer (RNFL) measurements obtained with optical coherence tomography (OCT2; Carl Zeiss Meditec, Dublin, California, USA) are predictive of the development of glaucomatous change. DESIGN: Cohort study. METHODS: Participants were recruited from the University of California, San Diego (UCSD) longitudinal Diagnostic Innovations in Glaucoma Study (DIGS). One eye was studied from each of 114 glaucoma suspects with normal standard automated perimetry (SAP) and OCT RNFL imaging at baseline. The cohort was divided into two groups based on the development of glaucomatous change (repeatable abnormal visual fields and/or a change in the stereophotographic appearance of the optic disk). Cox proportional hazards models were used to determine the predictive ability of OCT RNFL thickness measurements. RESULTS: Over a 4.2-year average follow-up period, 23 eyes (20%) developed glaucomatous changes and 91 (80%) did not. At baseline, thinner RNFL measurements, higher SAP pattern standard deviation (PSD), "glaucoma" stereophotograph assessment, and thinner central corneal thickness (CCT) were associated with the study endpoints in univariate analysis. After adjusting for age, intraocular pressure (IOP), CCT, and PSD in multivariate models, a 10 mum thinner average, superior and inferior RNFL at baseline was predictive of glaucomatous change [hazard ratio (95% CI); 1.51 (1.11 to 2.12), 1.57 (1.17 to 2.18), and 1.49, (1.19 to 1.91), respectively]. Results were consistent when stereophotographic assessment was included in multivariate analysis. CONCLUSIONS: Thinner OCT RNFL measurements at baseline were associated with development of glaucomatous change in glaucoma suspect eyes. RNFL thinning was an independent predictor of the glaucomatous change, even when adjusting for stereophotograph assessment, age, IOP, CCT, and PSD.