Temporal signal-to-noise ratio (tSNR) provides a crucial metric for accessing ability of the acquisition to detect small temporal fluctuations of the signal and therefore determining sensitivity of BOLD fMRI measurements. It has been shown that tSNR may vary dramatically between multiple runs of accelerated single-shot EPI acquisitions, introducing additional variability within and across sessions and therefore reducing sensitivity in fMRI studies [1]. Fast low-angle excitation echo-planar technique (FLEET) recently proposed for acquiring autocalibration signal (ACS) in accelerated EPI improves tSNR of the acquisition and tSNR continuity across the slices by providing robustness to subject motion and respiration related artifacts [2]. In this work we investigated the effect of FLEET-ACS on improving tSNR variation between multiple runs of accelerated EPI acquisitions.Five healthy volunteers (4F/1M, 30±8yo) were scanned on a whole-body 7T scanner (Siemens Healthcare, Erlangen, Germany) using a set of 7 single-shot gradient-echo EPI protocols – three with FLEET-ACS: acceleration factor R=2,3,4; and four with conventional ACS: R=2 (single-shot EPI-ACS, “SS”), and R=2,3,4 (standard segmented EPI-ACS, “MS”). Protocol parameter values were: TE/TR=25/2000 ms, FOV=192 mm, 39 slices, spatial resolution 2.0×2.0 mm, slice thickness 2.0 mm, α=67°. For FLEET-ACS protocols FLEET α=10° (with 5 preparation pulses). This set of 7 runs was repeated in a randomized order four times for three subjects with 60 measurements (t=60), and twice for two subjects with t=120. An additional experiment using an agar phantom was performed, in which three sets were acquired with t=120.Motion correction was performed for the human data (FSL, MCFLIRT, middle time point as a reference) and tSNR maps for each run were calculated after linear detrending. Noise-to-noise ratio (NNR) maps, defined as the pixelwise ratio of tSNR values, were computed between two halves of each run, NNR within (NNRw) and between each pair of runs aligned using FLIRT (FSL), NNR between (NNRb). Example slices of the tSNR maps calculated for FLEET-ACS and EPI-ACS acquisitions with R=2 (SS and MS) acquired for a human subject are presented in Figure 1. The highest tSNR values were observed for FLEET-ACS acquisition and EPI-ACS MS acquisition showed discontinuity of tSNR across the slices. Values of NNRw were relatively low and similar for different ACS techniques (Figure 2). NNRb values were generally higher than NNRw and lower for FLEET-ACS than for EPI-ACS acquisitions, as presented in Figure 3. In addition, NNRb maps calculated for EPI-ACS acquisitions suffered from spatial inhomogeneity effects, especially from slice discontinuity for the MS acquisitions. Similar discontinuity effect was also noticeable at the NNRb maps calculated for the phantom data (Figure 4).Our work confirmed a run-to-run tSNR variation previously reported for accelerated EPI acquisitions [1]. We showed that acquisitions using FLEET-ACS reach higher tSNR values and perform better in terms of run-to-run tSNR variation than the conventional EPI-ACS techniques. This is especially the case for the multi-shot EPI-ACS acquisitions strongly affected by the tSNR discontinuity between the slices (Figure 3). This artifact, as well as other spatial inhomogeneity effects were minimized in the SSNw maps calculated for FLEET-ACS.We showed that FLEET-ACS technique can reduce run-to-run variability of tSNR in accelerated EPI time-series data, increasing sensitivity of BOLD fMRI measurements. This reduction of variability across runs can have a direct effect on both single subject and group analysis.