In many chromatographic software weighted polynomial regression is used for calibration curves. Weighting is a useful mean to account for measurement error that may depend on the detector response value. Confidence intervals show, how accurate is a measurement with the help of a calibration curve. Besides, theory of confidence intervals was not extended to weighted polynomial regressions.

We extended confidence interval theory to the frequent case of weights, expressed as a function of Y value, in particular 1/Y and 1/Y². This extension allows accurate calculation of concentrations with confidence intervals for calibration curves, constructed using point weighting. Examples are shown that demonstrate applications of a weighted calibration curves with confidence intervals in chromatography.