This prospective study examined the etiology of eosinophilic pleural effusions investigated at a Thai thoracic center from January 1996 to February 1998. Among the 405 eligible pleural effusions, 31 were eosinophilic (EoPF) and 374 were noneosinophilic (NEoPF). Malignant effusions were established in 159 of the 405 patients, yielding a prevalence of 0.39. Malignant effusions were responsible in 24 of the 31 EoPF (77.4%), and 135 of the 374 NEoPF (36%)(p = 0.01). Bayesian analysis showed the post-test probability of malignancy in eosinophilic pleural effusions among our patient population to be 0.76. Tuberculous pleuritis was the etiology in 155 patients with NEoPF (41.4%) but in none of the patients with EoPF (p <0.001). There was no significant difference between EoPF and NEoPF in miscellaneous causes including paragonimiasis, amebiasis, lupus pleuritis, chylothorax, and yellow nail syndrome. It is concluded that eosinophilic pleural effusions are at least as likely to be malignant as noneosinophilic effusions. The finding of eosinophilic pleural effusions should not be regarded as suggestive of benign conditions.