In essential thrombocythemia patients alpha-IFN rapidly reduces platelet count, and it is also able to maintain a low count during long-term treatment. In order to verify if long-term IFN treatment can produce sustained remission in selected patients, we decided to suspend IFN treatment in two subsets of 21 patients on long-term alpha-IFN treatment: (a) all six patients who had shown a platelet count below 450 x 10(9)/l for at least 2 months with 3 MU once a week; (b) three patients who had shown the same platelet count for at least 2 months with 3 MU three times a week. After withdrawal of alpha-IFN treatment, a rapid increase in the platelet count was observed in all three patients requiring 3 MU three times a week. Three of the six patients treated with 3 MU once a week are still free of symptoms and have been in complete hematological remission (platelet count below 450 x 10(9)/l) for 9+, 13+, and 14+ months, respectively. As far as the three remaining cases are concerned, one was not assessable because of loss to follow-up, while the other two relapsed after 1 and 2 months. We believe that the three cases of sustained remission might be the result of a long-term tumor load reduction produced by the alpha-IFN treatment. Finally, the factor best able to predict sustained, unmaintained remission seems to be the clinical response to a low dose of alpha-IFN during the maintenance phase, rather than disease features prior to treatment.