There should be a recognisable trend between the incidence of hypereosinophilia and the duration of tropical endomyocardial fibrosis (EMF), if the hypothesis, that EMF is the burnt out phase of eosinophil associated heart disease, is correct. We tested this hypothesis in a prospective study of 89 consecutive EMF cases over an 18 year period at two Nigerian locations (Ife in South West and Calabar in South East). We carefully dated the duration of EMF symptoms at first presentation and screened for hypereosinophilia (eosinophilia > or = 1500/mm3), and their causes. When no cause was identified for hypereosinophilia we gave a therapeutic trial with diethylcarbamazine to the patients. An eosinophil count > or = 1000/mm3 was recorded in 80% of 24 cases seen within 6 months; 55% of 46 cases seen 0-24 months; 54% of 18 cases seen 25-48 months and 21% of 25 cases seen more than 49 months of onset of symptoms: while the respective distribution of eosinophil count > or = 1500/mm3 for similar periods were 66, 44, 27 and 21%. This reflects a highly significant (P < 0.001) inverse relationship between hypereosinophilia and the duration of EMF and strengthens the concept that EMF (without eosinophilia) represents the late stage of eosinophilic heart disease. The localisation of endemic EMF to the low-lying tropical rain forest Zone and its predominant occurrence among rural dwellers and farmers suggest a vector borne etiologic agent. Microfilaria was the most likely cause of hypereosinophilia in the cases presented.