CEST (Chemical exchange saturation transfer) acquisition is usually limited to single or several slice as acquisition at multiple spectral offsets as well as sufficiently long repetition time are needed. This suffices for studies where the region of interest is known a priori, such as for the cases of tumor. However, potential CEST applications in brain functional studies call for a 3D acquisition due to: 1) homogenous CEST saturation across slices; 2) intrinsically higher SNR; 3) better registration to other image contrasts. Previous attempts have been made using either 3D GRASE [1] and 3D EPI readout [2], and the overall acquisition time were still relatively long. In this work, CEST acquisition is implemented based on concentric stack-of-spiral, and the goal is to further improve the time per volume by exploiting the high efficiency of spiral acquisition. The proposed 3D CEST sequence is illustrated in Fig.1a. A non-spatial selective Fermi windowed RF with crusher gradients is used as CEST saturation, followed by 3D excitation. A FSE based stack of spiral readout is then played for data acquisition. All the slice encodings are placed within a single echo readout train, and multiple spiral arms are acquired in adjacent TRs. In this way, increasing the number of slices increase the ETL rather than the scan time and hence the impact of number of slices on the overall scan time is minimized, whereas the number of spiral arms can be adjusted for image resolution and overall scan time. A concentric readout (Fig.1b) in the z direction is used so that the first readout corresponds to the k-space center to give a near PD image contrast, which maximizes CEST saturation homogeneity and sensitivity.A APT experiment was performed using the proposed 3D CEST sequence on a patient diagnosed with glioma, consent form was obtained prior to the scan. A RF train consisting of 4 RFs of 400ms each was used as the CEST saturation. CEST acquisition covered a spectral range of 6ppm with a non-uniform density and a total of 39 points. B0 correction is made using the densely acquired spectral points around 0ppm.A 4 arm spiral trajectory was used with 1024 points along each arm [3], with a FOV of 220mm this leads to an analytical resolution of around 4mm. A total number of 38 slices with 4mm slice thickness were acquired. A TR of 2.5s was used with a minimum TE of 10.1 ms. The total scan time was 6:30 minutes for a full brain coverage. A spatially matching contrast enhanced 3D SPGR acquisition was also acquired for anatomical structure registration. The axial, sagittal and coronal slices of the acquired 3D volume (without RF saturation) are shown in Fig.2a. Relatively weak GM/WM contrast was seen as expected and. The map at 3.5ppm for the orthogonal slices are overlaid on the spatially matching slices from the 3D SPGR and shown in Fig.2b. It is seen that the high regions have good anatomical confirmation in all the slice. The 3D extent of the tumor was manually defined based on the contrast enhanced image, and the averaged z-spectrum of the lesion volume is shown in Fig.2c. It is seen that the level of at 3.5ppm shows good agreement with previously reported values. 3D acquisition is desired in many potential CEST applications but limited by practical scan time. Spiral is a highly efficient acquisition strategy and placing slice encoding along the echo train readout minimizes the impact of 3D encoding on the overall scan time. Keeping the same readout length, increasing the z direction coverage only increases the T2 decay related blurring. On the other hand, the in plane resolution may be improved by limiting the z coverage while lengthening the spiral readout. In this work, 10s per brain volume was achieved at a 4mm isotropic resolution, and further optimization of the acquisition strategy is possible by balancing desired SNR, resolution and spatial coverage. The relatively low spatial resolution and weak contrast due to PD weighting can be overcome by fusion image with anatomical image contrasts, as performed in other functional imaging such as arterial spin labeling.