The identification of an acquired mutation of JAK2 in patients with myeloproliferative disorders has raised questions about the relationship between mutation-positive and mutation-negative subtypes, timing of the JAK2 mutation, and molecular mechanisms of disease progression. Here we demonstrate that patients with V617F(-) essential thrombocythemia do not commonly progress to become V617F(+). Consistent with the concept of distinct pathogenetic mechanisms, we show that patients with and without the JAK2 mutation have different patterns of cytogenetic abnormality, with virtually all patients carrying the 20q deletion or trisomy 9 being V617F(+). We also investigated the existence of a "pre-JAK2" phase by comparing the proportion of clonally derived granulocytes, estimated from X-chromosome inactivation patterns (XCIPs), with the proportion of V617F(+) granulocytes. Our results demonstrate that inherent XCIP variability between granulocytes and T cells produces a systematically biased pattern of results that may be misinterpreted as evidence for an excess of clonally derived granulocytes, an observation that limits the utility of XCIP analysis in this context. Lastly, we studied 4 patients with V617F(+) myeloproliferative disorders who subsequently developed acute myeloid leukemia. In 3 patients the leukemic cells were V617F(-), suggesting that in these patients the leukemia arose in a V617F(-) cell.