Alpha/beta separation is achieved by the use of pulse shape discrimination (PSD), calibrated by quantifying event misclassification at any given setting for pure alpha and beta emitters. Previous studies have shown that the degree of misclassification is affected by quenching, but with no attempt made to understand the causes for this phenomenon. This study examines the potential effects of beta event energy and quench on PSD calibration. PSD was shown to be energy dependent, with misclassification increasing with beta event energy. Therefore, PSD calibration requires the use of a beta emitter with the same energy distribution as is present in the sample, or a restricted region of interest should be employed. For gross alpha/gross beta analysis of samples containing unknown beta emitters, a stepwise calibration procedure is proposed for both PSD and efficiency calibration. Quenching by carbon tetrachloride, nitromethane and 9 M hydrochloric acid was shown to affect PSD by suppression of the delayed component of the scintillation pulse, although to variable extents, and therefore having a range of effects on misclassification. Acetone quenching had little impact on PSD. Standard quench calibration procedures using either carbon tetrachloride or nitromethane are inappropriate for applications utilising alpha/beta separation by PSD. Instead, a quench calibration procedure based on overspiking samples encompassing a range of quench conditions is recommended. The influence of oxygen quenching on PSD was investigated through purging samples with either oxygen or nitrogen. PSD was found to be unaffected by the small amounts of oxygen normally dissolved in the diisopropylnaphthalene based cocktail used in this study.