Since all of noninvasive optical monitors such as near-infrared spectroscopy are transcutaneous, it is important to consider the optical properties of the skin. The purpose of the present study was to clarify the effects of incident angle and source-detector separation (SDS) on the detection of components in the dermal layer. Here, we developed a novel skin phantom consisting of epidermal and dermal layers with optical properties ranging from 400 to 1600 nm. The phantom was simulated by only water, scatters and absorbers, without agarose, silicone or other materials. The phantom showed reflectance spectra very similar to those of actual human skin by the integrating sphere measurements. Furthermore, an optical system was assembled in which the incident angle could be changed from 20º to 80º and the SDS from 0 to 6 mm independently. For the wavelength range of 400 to 900 nm, the absorption spectrum of hemoglobin in the dermal layer was investigated to assess detectability. For the wavelength range of 900 to 1600 nm, the absorption spectrum was confirmed by including glucose in the dermal layer. The absorbance was calculated from the measured diffuse reflected light intensity. The optimal incident angle and SDS for optical measurements focused on the dermal layer were estimated by the signal-to-noise ratio (SNR). In the wavelength range of 400 to 900 nm, the absorption spectrum of hemoglobin with the highest SNR was obtained at an incident angle of 70° and SDS of 4 mm. In the wavelength range of 900 to 1600 nm, the absorption spectrum of glucose with the highest SNR was obtained at an incident angle of 20° and SDS of 0 mm. These conditions are expected to be optimal for transcutaneous measurement of biomolecules within the dermis using diffuse reflected light of wavelength range from 400 to 1600 nm.