In an attempt to elucidate better the various aetiologies of pericardial effusion, we developed a diagnostic protocol that incorporated a battery of systematic tests including blood cultures, throat swab cultures and serological tests for various infectious agents and estimation of serum antinuclear antibodies and serum thyroid-stimulating hormone. Over a 2-y period ending May 2000, we evaluated prospectively and diagnostic usefulness of our strategy in a cohort (n = 136) of patients with pericardial effusion treated at Hospital Timone (HT), Marseille. We compared our findings with those observed in a retrospectively (May 1998-May 2000) drawn cohort (n = 127) of patients treated at Hospital Louis Pradel (HLP), Lyon and in which the laboratory investigation towards establishing an aetiological diagnosis was undertaken intuitively. Overall, the aetiologies were obvious clinically in 18% of cases. In other cases, specific aetiologies (27.3% vs 3.9%; p < 0.001), including treatable conditions (11.1% vs 2.4%; p < 0.001) were identified significantly more frequently in the HT cohort compared to the HLP cohort. The diagnosis strategy we propose may be helpful in elucidating the aetiological diagnosis of pericardial effusion when a cause is not obvious clinically.