Objective: To assess the sensitivity, specificity and predictive validity of suggested cut-off scores in the SF-36 mental health (MH) and mental component summary (MCS) in screening for depressive and anxiety disorders in a population sample of older Swedish women. Method: The sample comprised 586 randomly selected females aged 70–84 years who took part in an in-depth psychiatric examination. This provided the ‘gold standard’ against which the usefulness of SF-36 recommended thresholds for screening for depressive and anxiety disorders in older Swedish women was examined. Results: Based on DSM-III-R criteria, 69 women (12) were diagnosed with depression (major depression, dysthymia and/or depression NOS) and 49 (8) with generalised anxiety and panic disorders. The previously recommended MH and MCS cut-offs (i.e. 52 and 42) gave a specificity for diagnosis of depression of 92 and 82 and sensitivity of 58 and 71, respectively. Both the MH and MCS were good predictors of depressive disorders but poor predictors of anxiety disorders. Conclusion: The study supports the predictive validity of suggested SF-36 MH and MCS cut-off scores in screening for depressive disorder but not for anxiety disorder in older women in Sweden.