Estimates of the density spectrum of the macular pigment (Wyszecki G, Stiles WS. Color Science: Concepts and Methods. Quantitative Data and Formulas. 1st ed. New York: Wiley, 1967); (Vos JJ. Literature review of human macular absorption in the visible and its consequences for the cone receptor primaries. Institute for Perception. Soesterberg, The Netherlands, 1972) are partially based on the difference between central and peripheral spectral sensitivities, measured under conditions chosen to isolate a single cone class (Stiles WS. Madrid: Union Internationale de Physique Pure et Appliquée, 1953;1:65-103). Such derivations assume that the isolated spectral sensitivity is the same at both retinal locations, save for the intervening macular pigment. If this is true, then the type of cone class mediating detection should not influence the calculated difference spectrum. To test this assumption, we measured central and peripheral spectral sensitivities in a deuteranope, a protanope and a normal trichromat observer: (a) for short-wave sensitive (S-) cone detection; and (b) for long-wave sensitive (L-) cone detection (deuteranope), for middle-wave sensitive (M-) cone detection (protanope) or for both L- and M-cone detection (normal trichromat). The difference spectra determined for L- or M-cone detection deviate significantly from those measured for S-cone detection, at wavelengths below 450 nm. A theoretical analysis suggests that the discrepancies are owing, in part, to regional variation in the optical density of the cone pigments; and that such receptor variation cannot be ignored when deriving the standard density spectrum of the macular pigment.