The recently promulgated uncertainty theory of instrumental analyses w
as used to minimize the relative standard deviation of measurements. N
aphthalene, acenaphthene, pyrene and perylene in HPLC were used as ill
ustrations. For a Gaussian model peak with a width (S.D., sigma) of fi
fteen data points, the optimum integration domain was about +/-6 data
points (+/-0.4 sigma) around the peak centre on the HPLC baseline exam
ined. Two commonly used integration modes were examined: horizontal ze
ro line and oblique zero line. The precision was almost the same for b
oth modes in the HPLC analysis. The signal shape at the limit of detec
tion was also shown for the optimum integration domain. The error acco
mpanying the use of the uncertainty theory was evaluated using a Monte
Carlo simulation. The practical applicability and limitations of the
theory are discussed.