In Reply Li and colleagues raise a number of issues, which we address here. With respect to multiplicity, as stated in our article,1 we only interpreted P values over .001 as showing significant associations because of multiple comparisons. Examining the association of depression with cardiovascular disease and mortality was prespecified in our detailed protocol, as the Prospective Urban Rural Epidemiology (PURE) study was based on the results of the INTERHEART study2 and several previous studies, which had identified depression as an independent risk factor for myocardial infarction.3