Baseline adjustment is an important consideration in thorough QT/QTc studies for non-antiarrhythmic drugs. The time-matched baseline is often used in practice to adjust for the within-subject diurnal effects. More recently, the day-averaged baseline has been proposed in the literature to improve the efficiency of treatment comparisons for the primary and assay sensitivity hypotheses. We recommend an analysis of covariance (ANCOVA) model with change from the time-matched baseline as the response and both the time-matched baseline and the day-averaged baseline as the covariates. This model adjusts for the within-subject diurnal effects and at the same time is more efficient and robust than the ANCOVA models based on either the time-matched baseline alone or the day-averaged baseline alone. Copyright © 2012 John Wiley & Sons, Ltd.