AIMS: The main objective of this study was to investigate distributional shifts underlying observed age and cohort differences in mean levels of psychological distress in the 1958 and 1970 British birth cohorts. METHODS: This study used data from the 1958 and 1970 British birth cohorts (n = 24,707). Psychological distress was measured by the Malaise Inventory at ages 23, 33, 42 and 50 in the 1958 cohort and 26, 34, 42 and 46-48 in the 1970 cohort. RESULTS: The shifts in the distribution across age appear to be mainly due to changing proportion of those with moderate symptoms, except for midlife (age 42-50) when we observed polarisation in distress- an increase in proportions of people with no symptoms and multiple symptoms. The elevated levels of distress in the 1970 cohort, compared with the 1958 cohort, appeared to be due to an increase in the proportion of individuals with both moderate and high symptoms. For instance, at age 33/34 42.3% endorsed at least two symptoms in the 1970 cohort vs 24.7% in 1958, resulting in a shift in the entire distribution of distress towards the more severe end of the spectrum. CONCLUSIONS: Our study demonstrates the importance of studying not only mean levels of distress over time, but also the underlying shifts in its distribution. Due to the large dispersion of distress scores at any given measurement occasion, understanding the underlying distribution provides a more complete picture of population trends.