Background Little attention has been directed to the characteristics of electrocardiograms (ECGs) with brief QRS durations (BD). Methods From a database of 859 977 computer-analyzed (Marquette 12SL, GE Healthcare, Milwaukee, WI, USA) ECGs of 216 148 patients, 1805 patients (data set of BD: 19 718 ECGs; age, >18 years, unpaced), had, in at least one ECG, a QRS duration of less than 62 milliseconds, a prevalence of 0.8%. Results The QRS duration ranged from 46 to 188 milliseconds; values fluctuated between excessive brevity and the traditionally normal (80-95 milliseconds), with many values in the 60's. The average age at first acquisition was 56 years (range, 18-96 years); 1371 were female (76%) and 433 were males (24%); and 416 were white (23%), 1243 African American (69%), 109 other (6%). The summed 12-lead QRS amplitude (ΣQRS) ranged from 2800 to 32 929 μV (mean ± SD, 11 154 ± 4101), and heart rate ranged from 40 to 269 (98 ± 26). There was a statistically positive relationship between QRS duration (in the range 52-105 milliseconds) and ΣQRS ( P < 0001): for each 1000 μV increment, duration increased by 0.7 milliseconds (95% confidence interval [CI], 0.66-0.73 milliseconds). Conversely, for each 5-millisecond increment in duration, ΣQRS increased by 475 μV (95% CI, 450-498 μV). There was an inverse relationship between QRS duration and heart rate ( P < .0001): for every 10 beat rate increase, the duration dropped by 1.14 milliseconds (95% CI, 1.09-1.19 milliseconds). An additional data set of normal QRS durations ND consisted of 2902 subjects, none of whose 40 327 ECGs showed a QRS duration less than 62 milliseconds; 2.8% of ND ECGs had durations in the 62–69 millisecond range, 26% in BD. Sinus tachycardia was 43% in BD and 24% in ND. The average age of ND was 58 years (range, 18-100 years); 44% were male and 56% female; and 39% were white and 57% were African American. A positive relationship between duration and amplitude was found among ND subjects as well: for each 1000 point increase in ΣQRS, duration increased by 0.45 milliseconds ( P < .0001; 95% CI, 0.42-0.47 milliseconds). The duration amplitude relationship did not hold for durations greater than 105 milliseconds in either data set. Conclusion Over the full extended span of QRS durations (46-106 milliseconds), the value of the latter is directly related to the total summed amplitude of the 12-lead QRS. An inverse relationship is demonstrable with heart rate.