Background The longer term impact of flooding on health is poorly understood. In 2015, following widespread flooding in the UK during winter 2013/14, Public Health England launched the English National Study of Flooding and Health. The study identified a higher prevalence of probable psychological morbidity one year after exposure to flooding. We now report findings after two years. Methods In year two (2016), a self-assessment questionnaire including flooding-related exposures and validated instruments to screen for probable anxiety, depression and post-traumatic stress disorder (PTSD) was sent to all participants who consented to further follow-up. Participants exposure status was categorised according to responses in year one; we assessed for exposure to new episodes of flooding and continuing flood-related problems in respondents homes. We calculated the prevalence and odds ratio for each outcome by exposure group relative to unaffected participants, adjusting for confounders. We used the McNemar test to assess change in outcomes between year one and year two. Results In year two, 1064 (70%) people responded. The prevalence of probable psychological morbidity remained elevated amongst flooded participants [n = 339] (depression 10.6%, anxiety 13.6%, PTSD 24.5%) and disrupted participants [n = 512] (depression 4.1%, anxiety 6.4%, PTSD 8.9%), although these rates were reduced compared to year one. A greater reduction in anxiety 7.6% (95% confidence interval [CI] 4.6–9.9) was seen than depression 3.8% (95% CI 1.5–6.1) and PTSD: 6.6% (95% CI 3.9–9.2). Exposure to flooding was associated with a higher odds of anxiety (adjusted odds ratio [aOR] 5.2 95%, 95% CI 1.7–16.3) and depression (aOR 8.7, 95% CI 1.9–39.8) but not PTSD. Exposure to disruption caused by flooding was not significantly associated with probable psychological morbidity. Persistent damage in the home as a consequence of the original flooding event was reported by 119 participants (14%). The odds of probable psychological morbidity amongst flooded participants who reported persistent damage, compared with those who were unaffected, were significantly higher than the same comparison amongst flooded participants who did not report persistent damage. Conclusions This study shows a continuance of probable psychological morbidity at least two years following exposure to flooding. Commissioners and providers of health and social care services should be aware that the increased need in populations may be prolonged. Efforts to resolve persistent damage to homes may reduce the risk of probable psychological morbidity. Electronic supplementary material The online version of this article (10.1186/s12889-018-5236-9) contains supplementary material, which is available to authorized users.