OBJECTIVES: To determine (1) reference values for cone and rod phototransduction variables derived from the a-wave of the electroretinogram, (2) their dependence on age, (3) the progression in cone and rod variables in patients with X-linked retinitis pigmentosa (XLRP), and (4) the test-retest variability in these a-wave measures compared with the variability in cone and rod b-wave measures. PARTICIPANTS: One hundred control subjects aged 5 to 75 years and 24 patients with XLRP aged 5 to 38 years. METHODS: High-intensity stimuli were used to elicit electroretinograms in the dark and in the presence of a rod-saturating background. Computer averaging and computer subtraction of cone components from mixed rod-cone responses were used to derive rod-only and cone a-waves. Rod and cone phototransduction variables were derived by computer fitting physiologically based computational models to the leading edges of a-wave ensembles. RESULTS: Phototransduction efficiency, as indexed by the sensitivity variable (S), decreased with age for cone and rod-only responses, whereas maximum cone and rod photoresponses (Rm(P3)) remained constant. In patients with XLRP tested annually for 4 years, Rm(P3) for rods and, to a lesser extent, cones declined with disease progression, whereas S remained stable. The test-retest variability in the a-wave Rm(P3) is lower than previously reported measures of the variability in b-wave peak-to-peak amplitude. CONCLUSION: The leading edge of the a-wave of the electroretinogram can be related to rod and cone phototransduction variables through quantitative models. Rm(P3), rather than S, should be the outcome measure of choice when using the a-wave to follow photoreceptor function in prospective studies and treatment trials.