Boundary and landmark as visual spatial cues exert different effects in human spatial navigation. However, it is unclear that how different contributions between boundary and landmark are during human spatial navigation, and how their learning processes occur. In this study, we addressed these issues by using boundary-based and landmark-based spatial navigation tasks in a large sample of participants. During the task, participants were instructed to learn an object’s location based on boundary or landmark in the learning phase, and then retrieve the object’s location in the testing phase. Firstly, we found significantly lower distance and angular errors, and smaller variability during the boundary-based task than that in the landmark-based task, which suggested that boundary cue might guide more effectively and stably than landmark cue during spatial navigation. Secondly, our results showed that individual’s distance and angular errors declined less across the time in the boundary-based navigation than that in the landmark-based navigation, suggested that the boundary-based learning effect was weaker than the landmark-based learning effect. Finally, we found that boundary-influenced individuals were more proficient in the boundary-based navigation, while landmark-influenced individuals were more proficient in the landmark-based navigation. Together, these findings indicate that boundary has higher effectiveness for guiding but less enhancement for learning than landmark in spatial navigation.