© 2020 The Authors Background: This self-administered mail survey examined whether post-disaster residential location after the Fukushima nuclear power plant accident affected psychological distress, with related factors. Methods: We targeted 180,604 individuals, including 24,177 residents within evacuation order areas, in FY2011 and FY2015. Psychological distress trajectory was examined using a growth mixture model (GMM) on a total K6 score per year. Relationships between post-disaster residential location and psychological distress trajectory were examined. Results: GMM analysis showed five distress trajectories: Group 1, continuously low (less distressed) K6 scores from FY2011; Group 2, higher than Group 1 but lower than other groups; Group 3, similar to Group 2 in FY2011 but increasing afterwards; Group 4, gradual decline (improvement) from high scores in FY2011; Group 5, continuously high throughout. Post-disaster residential location in FY2011 and trajectory type was significantly associated after adjusting for covariates. Within-prefecture residents outnumbered out-of-prefecture residents in Group 1 and vice versa in Group 5. Significant associations between residential location and distress trajectory disappeared when additional covariates (job loss, FY2012 social network, and FY2012 problem drinking) were entered into multinomial logistic regression models. Similar tendencies were observed with FY2015 residential location as the exposure variable: those who continued living out-of-prefecture had higher psychological distress than those living within (or returned to) the prefecture, owing to covariates. Conclusions: To prepare for future disasters, networks and systems must be established to support problem drinkers and assist with re-employment. A seamless social network will allow evacuees access to appropriate support wherever they evacuate to.