BackgroundMeasurements of QRS duration (QRSD) in patients undergoing cardiac resynchronization therapy (CRT) are not standardized. We hypothesized that both the measurement of QRSD and its predictive value on CRT response are sensitive to the method by which QRSD is measured.MethodsElectrocardiograms (ECGs) pre‐ and post‐CRT from 52 CRT patients (66 ± 12 years, 65% male) were retrospectively analyzed. Custom‐made software was developed to measure global QRSD (QRSDglobal) and lead‐specific QRSD (QRSDI,II,III,aVR,aVL,aVF,V1,V2,V3,V4,V5,V6). QRSD was also assessed automatic by a routinely used ECG device. For each method we measured QRSD pre‐ and post‐CRT and shortening of QRSD (∆QRSD). Response to CRT at 6 months was defined as an improvement of ≥1 class in New York Heart Association classification and an increase by >7.5% in left ventricular ejection fraction.ResultsThe CRT response rate was 77%(n = 40). Different methods to measure QRSD show divergent nominal values before (median range 152—172 ms, P < 0.001) and after CRT (130–152 ms, P < 0.001). The predictive value of QRSD measurements for CRT response also varies significantly according to the method used (range AUC pre‐CRT QRSD 0.400–0.580, P < 0.05; AUC post‐CRT QRSD 0.447–0.768, P < 0.05; AUC ΔQRSD 0.540–0.858, P < 0.05). Global QRSD measurements revealed lower variability compared to lead‐specific QRSD.ConclusionDifferent methods to measure QRSD yield not only different nominal values but also influence the value of QRSD in predicting CRT response. Measuring QRSD by a global method can help to standardize QRSD measurements in future studies.