We found a way to increase the precision with which biomolecules present at concentrations below 10(-10) M can be quantified by fluorescence correlation spectroscopy (FCS). The effectiveness of the way was demonstrated experimentally by using a single-element aspheric objective lens, which was newly developed to reduce the cost of FCS instruments. In the first part of this paper, the relative standard deviation (RSD) of FCS-based concentration measurements is estimated theoretically by an analytical approximation assuming the detection volume profiles in FCS setups to be Gaussian and by molecular simulations in which more realistic profiles are calculated from physical parameters of the measurement setups. In a limit of infinitely bright molecules and zero background emission, the analytical approximation predicts that the RSD at a concentration is minimized when the mean number of molecules in a detection volume is approximately 0.5. A detection volume of the order of 10(-13) L thus gives smaller RSD values for concentrations from 10(-11) to 10(-10) M than does one of the order of 10(-15) L, which is widely used in FCS. This prediction is supported by the molecular simulations, taking into account the finite molecule brightness and background emission. In the second part of the paper, the RSD is evaluated experimentally with an FCS setup with a detection volume of 1.1 x 10(-13) L. The newly developed objective lens, serving as the bottom of the sample cell in this setup, has a large numerical aperture (0.9) without using immersion liquid. When a calibration line was made by 30-s FCS measurements of Cy3-labeled, 112-mer single-stranded DNA solutions, the RSD roughly agreed with the simulation result and was less than 0.1 for DNA concentrations from 2 x 10(-11) to 10(-10) M.