OBJECTIVE--To elucidate why different types of contrast appear in the left atrium during transoesophageal echocardiographic contrast studies. This should lead to a more uniform definition of true patent foramen ovale. BACKGROUND--The Valsalva manoeuvre and cough are routinely used to enhance right to left shunt for the detection of patent foramen ovale. No information is, however, available on the effect of these manoeuvres on the intrinsic echogenicity of blood in the left atrium. METHODS--30 consecutive patients referred for transoesophageal echocardiography were studied. Gain settings were relatively high so that no details were lost. The appearance of contrast during normal respiration, cough, and the Valsalva manoeuvre was looked for in the left atrium with and without venous injection of 10 ml of 5% dextrose. Frequency of contrast appearance in the left atrium was expressed as a percentage. Intensity of contrast, when present, was graded 1 (mild), 2 (moderate), or 3 (equal to right atrial contrast during injection). Timing was assessed in cardiac cycles after the end of respiratory manoeuvres. RESULTS--Left atrial contrast appeared as a "snowstorm" flowing from the right pulmonary veins towards the middle of the left atrium. It was present respectively with and without contrast injection in eight and five patients during normal respiration, in 15 and seven during a cough, and in 20 and 14 during the Valsalva manoeuvre. When present, the mean intensity of contrast was 1.0 during normal respiration, 1.4 during a cough, and 1.4 during the Valsalva manoeuvre. The mean delay of contrast appearance was 3-4 cycles after release of the Valsalva manoeuvre and after onset of cough. CONCLUSIONS--Respiratory manoeuvres frequently induce the transient appearance of mild to moderate contrast in the left atrium, most often independently of venous injections. Mild contrast was seen only with high gain settings. This contrast is likely to be related to transient stasis in the pulmonary circulation. In some cases peripheral venous injections of dextrose solution produced, without any respiratory manoeuvre, a similar contrast after the first four cardiac cycles of the right atrium filling, which is likely to represent recirculation of the injected bolus through the pulmonary capillary bed. Therefore respiratory manoeuvres should always be performed before contrast injections to allow better distinction between this background and true patent foramen ovale or pulmonary arteriovenous fistula.