A number of issues have remained unanswered in the design of "thorough QT"(TQT) studies. In this randomized, placebo-controlled, two-period crossover study in 20 healthy subjects, replicate electrocardiograms (ECGs) were recorded on a digital 12-lead Holter recorder, extracted in a core ECG laboratory, and interpreted manually by a cardiologist. The observed within-subject variability was slightly greater when time-matched baselines were employed than when predose baselines were employed, whereas the magnitude of the increase in QTc was similar for both. Moxifloxacin 400 mg was associated with an observed 7.5-12.5 ms increase in the mean placebo- and baseline-corrected QTc interval. A PK-QTc model estimated a 3.9 ms increase in the QTc interval for every 1,000 ng/ml increase in moxifloxacin concentration. The QTc increases associated with moxifloxacin support the appropriateness of its use as a positive control in TQT studies. This crossover study failed to justify the use of time-matched baselines rather than the less resource-intensive predose definition of baseline.