Childhood sexual abuse (CSA) is strongly associated with risk for major depression (MD) but the degree to which this association is causal remains uncertain.

We applied structural equation modeling using the Mplus program to 1493 longitudinally assessed female twins from the Virginia Adult Twin Study of Psychiatric and Substance Use Disorders.

Our model included (i) retrospective self- and co-twin reports on CSA, (ii) major potentially confounding covariates, (iii) assessment of lifetime history of MD at two separate interviews, and (iv) mood-congruent recall (implemented by allowing current depressive symptoms to predict reporting of CSA). In a model with only measurement error, CSA explained 9.6% of MD. Including four key covariates reduced the variance explained to 5.3%, with the largest effects found for parental loss and low parental warmth. Adding the effect of mood-congruent recall to a final well-fitting model reduced the percentage of variance explained in lifetime MD (LTMD) by CSA to 4.4%. In this model, current depressive symptoms significantly predicted recall of CSA.

In a model correcting for measurement error, confounding and the impact of mood-congruent recall, CSA remains substantially associated with the risk for LTMD in women. These findings strongly suggest, but do not prove, that this association is causal, and are consistent with previous results in this sample using a co-twin control design, but also indicate that more than half of the uncorrected CSA–MD association is probably not causal. Traumatic life experiences contribute substantially to the risk for LTMD.