Purpose To assess the signal composition of cone photoreceptors three-dimensionally in healthy retinas using adaptive optics optical coherence tomography (AO-OCT). Methods Study population. Twenty healthy eyes of ten subjects (age 23 to 67). Procedures. After routine ophthalmological assessments, eyes were examined using AO-OCT. Three-dimensional volumes were acquired at 2.5° and 6.5° foveal eccentricity in four main meridians (superior, nasal, inferior, temporal). Cone densities and signal compositions were investigated in four different planes: the cone inner segment outer segment junction (IS/OS), the cone outer segment combined with the IS/OS (ISOS+), the cone outer segment tips (COST) and full en-face plane (FEF) combining signals from all mentioned cone layers. Additionally, reliability of a simple semi-automated approach for assessment of cone density was tested. Main outcome measures. Cone density of IS/OS, IS/OS+, COST and FEF. Qualitative depiction and composition of each cone layer. Inter-rater agreement of cone density measurements. Results Mean overall cone density at all eccentricities was highest at the FEF plane (21.160/mm.sup.2 ), followed by COST (20.450/mm.sup.2 ), IS/OS+ (19.920/mm.sup.2) and IS/OS (19.530/mm.sup.2). The different meridians and eccentricities had a significant impact on cone density, with lower eccentricity resulting in higher cone densities (p[less than or equal to].001), which were highest at the nasal, then temporal, then inferior and then superior meridian. Depiction of the cone mosaic differed between all 4 layers regarding signal size and packing density. Therefore, different cone layers showed evident but not complete signal overlap. Using the semi-automated technique for counting of cone signals achieved high inter-rater reliability (ICC .99). Conclusions In healthy individuals qualitative and quantitative changes in cone signals are found not only in different eccentricities and meridians, but also within different photoreceptor layers. The variation between cone planes has to be considered when assessing the integrity of cone photoreceptors in healthy and diseased eyes using adaptive optics technology.