The period of urine collection used to measure excretion of catecholamines varies in epidemiological practice. We set out to compare overnight with 24 hour collection. Twenty-four subjects each collected urine for 24 hours, with the overnight urine being separately collected. The correlation of overnight and 24 hour catecholamines was highest when both measures were standardised for creatinine excretion and when creatinine excretion was adjusted for urine flow rate. The observed correlations were 0.74 for dopamine, 0.81 for noradrenaline and 0.54 for adrenaline. The use of overnight collections may therefore require a sample size up to 1.5 times as large (for noradrenaline) or 3.4 times as large (for adrenaline) to achieve the same power as with 24 hour collections. However, the figures given exaggerate the advantage of 24 hour collections if these incorporate measurement errors that are not present in overnight collections.