Dynamic collimation is an important dose reduction mechanism for helical CT scans, especially for modern wide-beam scanner models. Its implementation and efficacy need to be studied to optimize CT scan protocols and to reduce unnecessary patient dose.The purpose of this study is to evaluate dynamic beam collimation for modern wide-beam CT scanners with direct measurements and to estimate the efficacy for dose reduction.By using a linear-array solid state detector, primary x-ray beam coverage was measured for four CT scanner models: GE Revolution CT, Siemens Somatom Force, Philips iQon, and GE LightSpeed VCT. Supported independently from patient table, the detector remained stationary at the isocenter during helical scans. Data lines were recorded every 0.24 ms throughout one entire helical scan, with a spatial resolution of 0.8 mm along the craniocaudal direction. The measurements were repeated for various scan parameters related to dynamic collimation, including beam collimation width, pitch, rotation time, and scan length. The recorded beam coverage area was used as a surrogate to total primary dose, to model different dynamic collimation mechanisms. The directly measured total radiation range was compared to table travel distance and nominal scan length which equals to the ratio between DLP and CTDIvol. Equations to calculate the percentage dose reduction with dynamic collimation were derived for different mechanisms.Three different dynamic collimation mechanisms were revealed and related linear model parameters were reported for different helical scan parameters. The nominal scan length used to calculate DLP was shown to vary for different dynamic collimation mechanisms. For typical head and abdomen scans with nominal scan lengths of 17.5 cm and 25 cm, percentage dose reduction from dynamic collimation ranged from 2% to 32%.In conclusion, with direct measurements of primary x-ray beam coverage, dynamic collimation mechanisms and related dose reduction effects were characterized for four modern wide-beam CT scanners.