OBJECTIVE--To assess the interobserver variability between two observers from different echocardiographic laboratories. DESIGN--Two observers reviewed video recordings blinded to the other's diagnosis. In part I (n = 88), they determined interobserver variability for spontaneous echo contrast, left atrial thrombi, and appendage thrombi. No diagnostic criteria for thrombi were defined. In part II (n = 85), diagnostic criteria for thrombi were defined. RESULTS--Part I: Both observers agreed in diagnosing spontaneous echo contrast in 97%, left atrial thrombi in 90%, left atrial appendage thrombi in 94%. Part II: With predefined criteria no disagreement occurred in diagnosing left atrial thrombi. In the diagnosis of left atrial appendage thrombi both observers agreed in 89%. The mean diameters of the 10 thrombi on which the observers agreed were greater than of the nine appendage thrombi on which they disagreed. CONCLUSIONS--Interobserver variability in the diagnosis of spontaneous echo contrast is low. Defined criteria decrease interobserver variability for left atrial and appendage thrombi, although one third of the thrombi diagnosed by one observer were not confirmed by the other. Interobserver variability is high in the assessment of small structures (< 15 mm) within the left atrial appendage.