Objective: To investigate the impact of different numbers of readout points (Nread) on image quality in zero echo time (ZTE) imaging of the lumbar spine, and to determine the optimal Nread. Methods: Sixteen healthy volunteers were recruited and underwent coronal ZTE scans of the lumbar spine with three different Nread (128, 144, and 160), to obtain three sets of images (Nread-128, Nread-144, and Nread-160). The signal intensity of the cortical bone, muscle, and cancellous bone was measured in each set, and the standard deviation of subcutaneous fat was defined as noise to calculate the signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR). Two observers subjectively evaluated the three sets of images using a 4-point scale across four aspects: cortical bone depiction, anatomical structure clarity, perceived image noise, and overall image quality. Inter-observer agreement was assessed using weighted Kappa statistics. Repeated-measures ANOVA was used to compare SNR and CNR, and the Friedman test was used to compare subjective image quality scores, with Bonferroni correction applied for post-hoc analyses. Results: For the SNR in all measured regions (cortical bone, muscle, and cancellous bone) and the CNR between tissues (cortical bone-muscle, cortical-cancellous bone), the results consistently showed Nread-160 > Nread-144 > Nread-128 (P < 0.001). In terms of subjective evaluation, the scores for the Nread-160 set were significantly higher than those for the Nread-128 set (P < 0.001), while the difference between the Nread-160 set and the Nread-144 set was not statistically significant. (P > 0.05). Conclusion: In ZTE scanning of the lumbar spine, increasing Nread helps improve image quality. Nread-144 provides high SNR and CNR, and its subjective image quality scores are comparable to those of Nread-160. Considering both scanning time and image quality, 144 is recommended as the optimal Nread for ZTE imaging of the lumbar spine.