Magnetic susceptibility is an important tissue contrast¹. Quantitative susceptibility mapping² (QSM) is fast becoming a routine clinical tool in the evaluation and assessment of neurological diseases. Unfortunately, there is currently no established standard scanning protocol, and it is uncertain whether different acquisition methodologies alter the estimate of magnetic susceptibility. Here, we compare the deep gray matter susceptibility in healthy adults acquired from various QSM sequences using either unipolar or bipolar readout gradients, accelerated imaging or not (i.e., ASSET R2 or R1, respectively), with or without gradient-warp correction.Four healthy adult volunteers were scanned three times each within four days at 3T (Discovery 750; GE Healthcare, Waukesha, WI) using a 12-channel, receive-only head and neck phased-array coil. The eight different QSM combinations (all of which are 3D, multi-echo gradient-echo, with 1.0x1.0x1.0 mm³ voxels) were acquired in different randomized order for each session, co-registered to an anatomical atlas, and referenced to cerebrospinal fluid. The average and standard deviation (SD) susceptibilities in the caudate, putamen, red nucleus, internal and external globus pallidi were used in a linear mixed effects model³ (LME) to determine the influence of the various acquisition parameters. The fixed effects are the readout gradient polarity, acceleration factor, and gradient-warp correction, whereas the random effects are the repeats and tissues, both grouped by subject.We had 480 observations (4 subjects, 3 repeats, 8 QSM variants, 5 tissues) and excluded 21 outliers such that we used 459 samples in the LME analysis. The initial LME models of the average and SD of susceptibility (Tables 1 and 2, respectively) used all four subjects and included all fixed effects and their interactions, along with all random effects and their interactions. The intercept denotes the internally defined reference (subject 1, repeat 1, caudate, unipolar readout gradient, non-accelerated, gradient-warp correction on). The very small p­-values indicate that the average and SD susceptibility are significantly different than zero. The only other statistically significant fixed effects coefficient was gradient-warp correction (GW_Off). To increase the power of our analysis, we pooled all of the non-significant effects accordingly and defined simpler LME models that included only the intercept, gradient-warp correction and the random effects (Tables 3 and 4). Likelihood ratio tests (p-values of 0.21 and 0.57 for the average and SD analyses, respectively) showed that the simpler LME models are not statistically different than their more complicated counterparts, so the simpler LME models are preferred. From the estimates in Tables 3 and 4, gradient-warp correction off increases the average and SD susceptibility by 5.5% and 2.8%, respectively.In the deep gray matter structures of healthy adults, readout gradient polarity and accelerated parallel imaging do not alter the susceptibility estimate. With gradient-warp correction on, the geometric fidelity is higher, but the average and SD estimates of susceptibility were found to be 3-5% lower.