Bicycle helmets attenuate head impacts using expanded polystyrene (EPS) foam liners. The EPS density plays a key role in determining the helmet and head response during an impact. Prior pilot work in our lab showed that EPS density varied by up to 18 kg/m3 within a single helmet, and thus the purpose of this study was to quantify the regional density variations within and between helmets and to establish how these variations influence helmet impact performance. We evaluated 10–12 samples of two traditional and two bicycle motocross (BMX) bicycle helmets with EPS liners. The bulk liner density and density of 16–19 cores extracted from specific locations on each sample were measured. Additional samples of two of these helmet models were then impacted at 3.0, 6.3, and 7.8 m/s to determine the relationship between local EPS density and helmet impact performance. We found that density varied significantly within each sample in all helmet models and also varied significantly between samples in three helmet models. The density variations were not symmetric across the midline in two of the four helmet models. The observed density variations influenced the helmets' impact performance. Our data suggest that variations in peak headform acceleration during impacts to the same location on different samples of the same helmet model can be partially explained by density differences between helmet samples. These density variations and resulting impact performance differences may play a role in a helmet's ability to mitigate head injury.