We are in full agreement with Carroll about the limited utility of clinical symptoms for ‘diagnostic tests’ and the consequent importance of efforts to discover biomarkers, endophenotypes or genetic markers. In fact, the main focus of our research is molecular genetic epidemiological investigation of mood disorders and psychoses that has precisely this aim. ^{Reference Green, Raybould, Macgregor, Gordon-Smith, Heron, Hyde, Grozeva, Hamshere, Williams, Owen, O'Donovan, Jones, Jones, Kirov and Craddock1–Reference Craddock, Jones, Jones, Kirov, Green, Grozeva, Moskvina, Nikolov, Hamshere, Vukcevic, Caesar, Gordon-Smith, Fraser, Russell, Norton, Breen, St Clair, Collier, Young, Ferrier, Farmer, McGuffin, Holmans, Donnelly, Owen and O'Donovan4} Further, we have a keen interest in using findings to provide biological validators for psychiatric nosology, classification and clinical diagnosis. ^{Reference Craddock and Owen5}

However, for the moment, psychiatrists have to make do with the clinical tools available and be alert to diagnostic clues that can help in the delivery of optimal care to their patients. We stand by the statements in our paper: ‘It is commonly, but wrongly, assumed that there are no important differences in the clinical presentation of unipolar and bipolar depression… The clinical features of depression are not, of course, a definitive guide to diagnosis but can help alert the clinician to a possible bipolar course… This is important because optimal management varies between bipolar and unipolar depression.’