Fluctuations in platelet count levels over time may help distinguish immune thrombocytopenia (ITP) from other causes of thrombocytopenia. We derived the platelet variability (PVI) score to capture both the fluctuations in platelet count measurements and the severity of the thrombocytopenia over time. Raw PVI values, ranging from negative (less severe thrombocytopenia and/or low fluctuations) to positive (more severe thrombocytopenia and/or high fluctuations) were converted to an ordinal PVI score, from 0 - 6. We evaluated performance characteristics of the PVI score for consecutive adults with thrombocytopenia from the McMaster ITP Registry. We defined patients with definite ITP as those who achieved a platelet count response after treatment with intravenous immune globulin or high dose corticosteroids; and possible ITP as those who never received ITP treatment or did not respond to treatment. Of 841 thrombocytopenic patients, 104 had definite ITP, 398 had possible ITP, and 339 had non-ITP thrombocytopenia. The median PVI score was 5 (interquartile range [IQR] 5, 6) for definite ITP; 3 (1, 5) for possible ITP; and 0 (0, 2) for non-ITP. A high PVI score correlated with the diagnosis of definite ITP even when calculated at the patient's initial assessment, before any treatment had been administered. Platelet count fluctuations alone contributed to the specificity of the overall PVI score. The PVI score may help clinicians diagnose ITP among patients with thrombocytopenia.