A parallel transmission (pTx) was developed to control the distribution of B1+, especially for ultrahigh field MRI. This study aimed to compare the intensity inhomogeneity of B1+ map and gradient echo images between two-channel pTx and quadrature transmission (qTx) at 7 Tesla (7T).A spherical water phantom and six healthy volunteers were scanned by 7T MRI using a two-channel quadrature head coil. A 500-mL water pack on the phantom created an artificial B1+ inhomogeneity. The B1+ map by the Bloch–Siegert method (1), two-dimensional-gradient-recalled-acquisition in steady state (GRASS), and 2D-spoiled-gradient-echo (SPGR) images were obtained with the pTx and qTx. Instead of RF shimming, we used the “brute force method” to obtain an optimal amplitude and phase for each channel of pTx. The simulated B1+ map of pTx was created using various combinations of amplitude and phase, or seventy-two different patterns (i.e., 8 patterns of relative phase × 9 patterns of relative amplitude) applied for each iteration. The next iteration based on the combination of minimum standard deviation of B1+map was conducted by narrowing the range of the phase and amplitude. The inhomogeneity of all images were evaluated with region-of-interest measurement using coefficient of variation (CV) values.The B1+ map of the phantom and volunteers showed better signal homogeneity with pTx than with qTx. The GRASS images of the phantom and volunteers were more homogeneous with pTx than with qTx, accompanying the B1+ improvement; however, the SPGR images showed no improvement with the pTx (Fig. 1 and 2). The mean CVs of the ROI analysis of pTx and qTx were 0.240 and 0.286, respectively, in the phantom B1+ map (p < 0.01); 0.508 and 0.567, respectively, in the phantom GRASS (p < 0.01); 0.185 and 0.192, respectively, in the volunteer B1+ map (p < 0.05); and 0.264 and 0.313, respectively, in the volunteer GRASS (p < 0.05). In contrast, the mean CVs of phantom and volunteer of pTx and qTx in SPGR were 0.315 vs. 0.316, and 0.097 vs. 0.096, respectively, which was not statistically different (p = 0.213 and 0.753), respectively (Fig.3 and 4).Inhomogeneity in the transmit RF field was not a serious problem before the advent of the ultrahigh field, such as 3T or 7T. In the ultrahigh field, the larger resonant frequency and the shorter RF pulse cause a strong B1+ inhomogeneity (2); therefore, methods for improving the B1+ inhomogeneity are needed. Previous studies (3-5) demonstrated the superiority of pTx over qTx similarly to our study, however most of these studies used a human head model (digital) phantom or few volunteers, and only focused on B1+ homogeneity. On the other hand, our study evaluated the B1+ map and the clinical gradient echo images by ROI analysis in both phantom and volunteers. Our pTx systems demonstrated B1+ map and GRASS images improvement by pTx comparing to qTx, however the signal inhomogeneity of phantom and volunteers’ heads was not fully controlled, perhaps due to the smaller number of RF channels of pTx (i.e. two channels), the influence of receiver sensitivity, and the RF pulse modulation type which could control only the amplitude and phase, not control the waveform of RF pulse. Additionally the SPGR images by pTx and qTx were not different in both phantom and volunteers. The discrepancies between GRASS and SPGR images were not clear, and may be because of the difference between flip angle and repetition time and not being susceptible to B1+ differences.The pTx system could improve the intensity homogeneity of B1+ map and GRASS images in phantom and the human brain, which could improve the clinical image quality of 7T MRI.