This study examines the physical, social, and temporal dimensions that affect Iranian women's perceptions of safety during travel on public transit. We employ both primary and secondary data: a survey of 680 female users of the Shiraz subway and GIS shapefiles. Our Partial Least Squares Structural Equation Modelling (PLS-SEM) explains 34.4 % of the variance in perceived safety. We find that constructs such as access, demographics, design, and surveillance have a direct effect on perceived safety, whereas the effect of timing is only indirect. The effect of surveillance measures, such as CCTV cameras, police officers, and lighting, is the most influential, explaining 43 % of the variance in perceived safety. This suggests that, in the case of Shiraz, enhancing these measures may be a particularly effective strategy to boost safety perceptions among female passengers. However, we acknowledge that excessive surveillance can backfire and may also be objectionable on ethical grounds. Beyond infrastructure measures, the real challenge is to cultivate a feeling of collective responsibility for women's and girls' safety and empower female riders to confront and report perpetrators.