Introduction: Assessment of bleeding in ITP is difficult. There is no agreed upon methodology; the WHO bleeding score does not seem useful for ITP. We report an ITP bleeding score including the relationship between site-specific bleeding severity (BLS), platelet count (plt ct), and plt size.Methods: A bleeding score was created which has 11 sites, 9 by history (Hx) of the previous week: skin, oral, epistaxis, GI, GU, GYN, hemoptysis, subconjunctival (SC), and intracranial hemorrhage (ICH); and 2 by Exam (PE): skin and oral. The BLS grades (gr) pts for each site on a defined scale from no bleeding (0) to “major” bleeding (2). Plt cts and large plts (LP) were measured on the Bayer Advia 120. No pt had > 2 study visits.Results: The 65 ITP pts on 100 visits over 6 months had a median (mdn) plt ct of 41,500/uL (41.5k). The mdn age was 31.5 yrs; 20 pts were children. 56 pts had chronic ITP, 26 s/p splenectomy, with a mdn duration of ITP of 6 yrs. There were no GU, SC hemorrhage (hem), or ICH. 72% of all gr 1 hem's were skin whereas 46% of gr 2 hem's were oral. There was 1 GI hem and 3 hemoptysis visits that were gr 2. 42 women on 63 visits had 5 visits each with GYN BLS gr 1 and gr 2. X2 testing correlated plt ct and BLS for 6 sites: both oral and both skin sites, epistaxis, and GYN (p's < 0.05). Oral PE correlated best with the plt ct (fig 1). For oral PE, BLS of gr 2 did not occur at plt cts > 20k.When the BLS was scored independently by two observers (LKP & JB) on 63 of the100 visits, 92% of individual BLS grs were identical (PearsonR=0.75). Skin PE and epistaxis each had 11/63 discrepancies. In addition, 16 BLS for oral (15 for skin) were different comparing Hx and PE (fig 2).LP did not correlate well with site-specific BLS when only plt cts ≤ 30k were included, suggesting that large, young platelets may not primarily determine bleeding severity in ITP. However for Skin Hx, if the BLS was gr 0, 4/6 had LP ≥ 2 whereas only 7/28 LP were ≥ 2 when the BLS was gr 1.Conclusions: The Bleeding Score reported here was validated by: a) the relationship of bleeding severity at multiple sites to plt ct including the absence of gr 2 bleeding at plt cts >20k; and b) the high degree of interobserver reliability despite a low but irreducible error rate based on variability in pt recall. It takes ≤ 3–5 minutes to complete. The significance of LP remains unclear. Studies of the BLS in larger numbers of patients will further determine its usefulness. [Display omitted] [Display omitted]