Although in EPI-fMRI analyses typical acquisition parameters (TR, TE, matrix, slice thickness, etc.) are generally employed, various readout bandwidth (BW) values are used as a function of gradients characteristics of the MR scanner. Echo spacing (ES) is another fundamental parameter of EPI-fMRI acquisition sequences but the employed ES value is not usually reported in fMRI studies. In the present work, the authors investigated the effect of ES and BW on basic performances of EPI-fMRI sequences in terms of temporal stability and overall image quality of time series acquisition. EPI-fMRI acquisitions of the same water phantom were performed using two clinical MR scanner systems (scanners A and B) with different gradient characteristics and functional designs of radiofrequency coils. For both scanners, the employed ES values ranged from 0.75 to 1.33 ms. The used BW values ranged from 125.0 to 250.0 kHz/64pixels and from 78.1 to 185.2 kHz/64pixels for scanners A and B, respectively. The temporal stability of EPI-fMRI sequence was assessed measuring the signal-to-fluctuation noise ratio (SFNR) and signal drift (DR), while the overall image quality was assessed evaluating the signal-to-noise ratio (SNR(ts)) and nonuniformity (NU(ts)) of the time series acquisition. For both scanners, no significant effect of ES and BW on signal drift was revealed. The SFNR, NU(ts) and SNR(ts) values of scanner A did not significantly vary with ES. On the other hand, the SFNR, NU(ts), and SNR(ts) values of scanner B significantly varied with ES. SFNR (5.8%) and SNR(ts) (5.9%) increased with increasing ES. SFNR (25% scanner A, 32% scanner B) and SNR(ts) (26.2% scanner A, 30.1% scanner B) values of both scanners significantly decreased with increasing BW. NU(ts) values of scanners A and B were less than 3% for all BW and ES values. Nonetheless, scanner A was characterized by a significant upward trend (3% percentage of variation) of time series nonuniformity with increasing BW while NU(ts) of scanner B significantly increased (19% percentage of variation) with increasing ES. Temporal stability (SFNR and DR) and overall image quality (NU(ts) and SNR(ts)) of EPI-fMRI time series can significantly vary with echo spacing and readout bandwidth. The specific pattern of variation may depend on the performance of each single MR scanner system in terms of gradients characteristics, EPI sequence calibrations (eddy currents, shimming, etc.), and functional design of radiofrequency coil. Our results indicate that the employment of low BW improves not only the signal-to-noise ratio of EPI-fMRI time series but also the temporal stability of functional acquisitions. The use of minimum ES values is not entirely advantageous when the MR scanner system is characterized by gradients with low performances and suboptimal EPI sequence calibration. Since differences in basic performances of MR scanner system are potential source of variability for fMRI activation, phantom measurements of SFNR, DR, NU(ts), and SNR(ts) can be executed before subjects acquisitions to monitor the stability of MR scanner performances in clinical group comparison and longitudinal studies.