Accurate length measurements by intravascular ultrasound (IVUS) are necessary for stent length selection and for IVUS volumetric analysis. The comparative accuracy of commercially available transducer pullback systems--a necessity for accurate IVUS length and volume measurements--has never been studied. We evaluated the accuracy of four IVUS pullback systems by studying 180 patients (45 in each group) who had been treated with a single stent of known length. Stented lesions were located in the left anterior descending artery (n = 77), left circumflex artery (n = 41), right coronary artery (n = 41), left main (n = 2), and saphenous vein grafts (n = 19). The known lengths of implanted stents ranged from 8 to 33 mm. The correlations between known stent length and IVUS-measured stent length in each group were 0.92 for CVIS, 0.83 for BSC Galaxy, 0.63 for Endosonics TrackBack, and 0.69 for Volcano Model R-l00 research pullback device, respectively. Furthermore, the absolute value of the difference between the two measurements was 9.1 +/- 13.1%, 8.8 +/- 10.2%, 18.6 +/- 21.5%, and 17.5 +/- 31.4%, respectively. With the Volcano Model R-l00 research pullback device, there were 3 extreme outliers; if these three outliers were excluded, then the correlation improved from 0.69 to 0.91; and the absolute deviation from known stent length improved from 17.5 +/- 31.4% to 9.7 +/- 8.3%. Thus, there is a significant variation in length measurement accuracy among IVUS pullback devices. This should be taken into account both clinically and when planning scientific studies.