Higher neighbourhood walkability would be expected to contribute to better health, but the relevant evidence is inconsistent. This may be because residents' dietary attributes, which vary with socio-economic status (SES) and influence their health, can be related to walkability. We examined associations of walkability with dietary attributes and potential effect modification by area-level SES.The exposure variable of this cross-sectional study was neighbourhood walkability, calculated using residential density, intersection density and destination density within 1-km street-network buffer around each participant's residence. The outcome variables were dietary patterns (Western, prudent and mixed) and total dietary energy intake, derived from a FFQ. Main and interaction effects with area-level SES were estimated using two-level linear regression models.Participants were from all states and territories in Australia.The analytical sample included 3590 participants (54 % women, age range 34 to 86).Walkability was not associated with dietary attributes in the whole sample. However, we found interaction effects of walkability and area-level SES on Western diet scores ( < 0·001) and total energy intake ( = 0·012). In low SES areas, higher walkability was associated with higher Western dietary patterns ( = 0·062) and higher total energy intake ( = 0·066). In high SES areas, higher walkability was associated with lower Western diet scores ( = 0·021) and lower total energy intake ( = 0·058).PHigher walkability may not be necessarily conducive to better health in socio-economically disadvantaged areas. Public health initiatives to enhance neighbourhood walkability need to consider food environments and socio-economic contexts.