In our recently published meta-analysis (Arts et al. Reference Arts2008) an error was made confusing s.e. (standard error) with s.d. (standard deviation) in the papers of Sobczak et al. (Reference Sobczak2003) and Nehra et al. (Reference Nehra2006) leading to an overestimation of effect sizes, especially regarding verbal memory in relatives. Our conclusion that ‘executive function and verbal memory are candidate bipolar endophenotypes’ has to be changed to ‘executive function is a candidate bipolar endophenotype’, because the effect of verbal memory in first-degree relatives is no longer significant. The general conclusion that effect sizes of cognitive deficits in relatives of patients with bipolar disorder are small continues to be valid.

The amended Results section and Tables 1–4 of the paper are reproduced in full and are available as Supplementary material, available online on the Journal's website (http://journals.cambridge.org/psm).

The authors apologize for these errors.