SS-FSE is a robust and fast method to acquire images in regions of $$$B_0$$$ inhomogeneity, unlike an EPI fast imaging sequence which presents geometric image distortions in such regions. However, in order to maintain signal throughout it’s long refocusing train, SS-FSE must generally maintain the CPMG condition [1]. This becomes problematic for sequences which introduce diffusion weighting after the initial $$$90^\circ$$$ as the excitation phase will be disturbed due to eddy current effects and motion, which subsequently causes spatially varying oscillation and rapid decay in the echo signal leading to image banding and ghosting. Previous efforts [2-3] have been made to overcome these issues, but each technique suffers from trade-offs. A full signal can be recovered by the technique in [3] (nCPMG), but this quadratic phase cycling requires that the refocusing flip angle is above $$$120^\circ$$$. To achieve a flip angle uniform across a slice, a more selective refocusing pulse is ideal, but this comes with a longer pulse duration and ultimately longer ESP, which causes $$$T_2$$$ modulation (i.e., blurring). We examine these effects for pulses with short as well as broad transition bands through simulation as well as phantom data on a SS-FSE sequence using nCPMG phase cycling. We propose a compromise in the form a recently proposed [4] DIVERSE pulse with a similar slice profile as the higher TBW pulse while achieving a suitable ESP for SS-FSE.A Bloch simulation was implemented to simulate the response of 100,000 spin isochromats placed over a 7.5cm spacing. This included a uniform $$$90^\circ$$$ excitation over a 5cm slice but with variable phase to demonstrate the effects when the excitation varies between being in phase and in quadrature. Slice profile effects from the refocusing pulses were included as described in [5]. Here two scenarios were simulated: (1) a case where the refocusing pulse was a short hamming-windowed sinc pulse with TBW=1.2 and (2) a TBW=3.5 SLR pulse. This SS-FSE sequence was implemented on a 1.5T GE scanner. Resolution phantom images were acquired using a 8ch cardiac coil using FOV=20cm, 192x128 matrix, TR=1500ms. In vivo images were acquired using a FOV=48cm. TE depended on ESP, which was dictated by the pulse type. Three scenarios were tested: (1) a low TBW=1.2 windowed sinc pulse (ESP = 4.1ms) , (2) a TBW=3.5 SLR pulse (ESP = 6.5ms) , and (3) a TBW=3.5 DIVERSE pulse (ESP = 4.9ms). Images were acquired and reconstructed using a hybrid ESPIRiT/homodyne reconstruction as described in [6]. Simulation results in Fig. (2.a) show results from high and low TBW pulses using this phase cycling scheme. The lower TBW pulses show higher signal, but suggest less signal stability based on the apparent signal modulation. Fig. (2.b) shows echo amplitude variations that corroborates the simulation in terms of signal level and stability. In all cases there is some signal modulation, which is manifest in image aliasing when reconstructed. With nCPMG phase cycling, the high TBW pulse had the most stable signal followed by the DIVERSE and then low TBW pulse. Fig. (2) shows reconstructed phantom images. The high TBW pulse performed the best by having the minimum difference between the CPMG and nCPMG cases. The DIVERSE pulse shows more aliasing and the low TBW pulse images show even more aliasing. The low TBW pulse provided the least phase encode blurring (seen on the left and right edges), while the DIVERSE followed closely with the high TBW pulse showing much more blurring. Fig. (4) shows in vivo data from a healthy volunteer. This data also supports the data from the phantom data where image sharpness (likely from both sharper slice profile as well as shorter ESP) is best with the DIVERSE pulse.We have simulated and demonstrated in phantom and in vivo studies the effects of various refocusing pulses on the signal stability and image quality in nCPMG SS-FSE acquisitions. We have seen that while a high TBW SLR pulse gives the most stable signal with the least aliased image, the longer ESP leads to phase encode blurring. A good compromise, then, is to use a DIVERSE pulse. We postulate that increased signal instability with the DIVERSE pulse is due to eddy currents and gradient RF system imperfections, and further work will include modifying a DIVERSE that is tuned for a nCPMG application, specifically reducing gradient slew requirements in order to minimize the impact from eddy currents.