To prospectively evaluate breath-hold capability and patterns, coronary artery rest periods, and beta-blocker use in coronary magnetic resonance (MR) imaging. Ethics committee approval and informed consent were obtained. In 210 consecutive patients (mean age, 61.8 years +/- 10.3 [standard deviation]; 146 men, 64 women), breath-hold patterns and maximal capability were assessed at expiration with dynamic navigator MR imaging (temporal resolution, 1 second). Left coronary artery (LCA) and right coronary artery (RCA) rest periods were determined at transverse cine imaging (steady-state free precession, retrospective gating, 40 phases per cycle). Before and after beta-blockade, rest periods were assessed in 25 additional patients (mean age, 61.4 years +/- 7.1; 20 men, five women). Differences were tested within groups with paired Student t test and between groups with unpaired Student t test (continuous variables) and chi(2) test (categoric variables). Pearson correlation was used to test the relationship between rest period and heart rate. Four distinct breath-hold patterns, characterized by diaphragmatic motion, were identified: pattern 1, steady plateau (55% of patients); 2, initial drift followed by plateau (12%); 3, continuous drift (19%); and 4, irregular, unsteady behavior (14%). Mean breath-hold capability with patterns 1 and 2 was 29 seconds +/- 13 (range, 10-64 seconds). The rest period of LCA was longer than that of RCA (163 msec +/- 75 vs 123 msec +/- 60; P < .01) and began earlier in the cardiac cycle (521 msec +/- 149 vs 540 msec +/- 160; P < .01); In a minority of patients, LCA rest period began later (21%) or was shorter (14%). With no pharmacologic intervention, correlation between rest period duration and heart rate was weak (LCA, r = -0.52; RCA, r = -0.38; P < .01). However, beta-blockade significantly lowered heart rate (61.3 beats/min +/- 7.2 vs 82.6 beats/min +/- 12.5, P < .001) and increased rest duration (LCA, 201.8 msec +/- 83.6 vs 111.8 msec +/- 44.55; RCA, 134.8 msec +/- 57.3 vs 83.1 msec +/- 35.8; P < .001). In 33% of patients (patterns 3 and 4), breath-hold pattern was unsuitable for high-spatial-resolution breath-hold MR imaging. LCA and RCA rest periods showed large variability in starting point and duration, with no correlation to heart rate.