To compare measured radiation dose (MD), estimated radiation dose (ED) and image quality in coronary computed tomography between turbo-flash (TFP) and retrospective protocol (RP) and correlate MD with size-specific dose estimates (SSDE).In this prospective study, we selected 68 patients (mean age, 59.2 ± 9.7 years) undergoing 192 × 2 dual-source CT (SOMATOM Force, Siemens) to rule out coronary artery disease. Thirty-one underwent TFP and 37 RP. To evaluate in vivo MD, thermoluminescent dosimeters were placed, superficially, at thyroid and heart level, left breast areola and left hemi-thorax. MD in each site, and ED parameters, such as volume CT dose index (CTDIvol), SSDE, dose length product (DLP), effective dose (E), were compared between two protocols with a t test. Image quality was compared between two protocols. Inter-observer agreement was evaluated with a kappa coefficient (k). In each protocol, MD was correlated with SSDE using a Pearson coefficient (r).Comparing TFP and RP, MD at thyroid (1.43 vs. 2.58 mGy; p = 0.0408), heart (3.58 vs. 28.72 mGy; p < 0.0001), left breast areola (3.00 vs. 24.21 mGy; p < 0.0001) and left hemi-thorax (2.68 vs. 24.03 mGy; p < 0.0001), CTDIvol, SSDE, DLP and E were significantly lower. Differences in image quality were not statistically significant. Inter-observer agreement was good (k = 0.796) in TFP and very good (k = 0.817) in RP. MD and SSDE excellently correlated with TFP (r = 0.9298, p < 0.0001) and RP (r = 0.9753, p < 0.0001).With TFP, MD, CTDIvol, SSDE, DLP and E were significantly lower, than with RP. Image quality was similar between two protocols. MD correlated excellently with SSDE in each protocol.