High plasma levels of the shed form of L-selectin (sL-selectin) are frequently detectable in acute myeloid leukemia (AML). sL-selectin can inhibit blast cell adhesion to vascular endothelium and may thereby influence the phenotype of AML. In this study, we have investigated the relationship between sL-selectin levels and clinical presentation or disease outcome in 100 patients with AML. Fifty-eight patients were found to have sL-selectin levels >/=3.12 microgram/mL (>/=3 SD above the mean of healthy controls: "increased"). Patients with extramedullary disease such as lymphadenopathies, splenomegaly, hepatomegaly, and/or muco-cutaneous infiltration had significantly increased sL-selectin levels (P < .001). sL-selectin levels were significantly heterogeneous in the French-American-British subtypes (P = .0003). Patients with "normal" sL-selectin levels had higher probability of achieving complete remission (CR) than with "increased" levels: 81% versus 64%, respectively (P = .06). When adjusting for clinically relevant covariates predictive for CR (sex, age, Auer rods), "normal" sL-selectin levels were significantly associated with CR (odds ratio, 3.08; 95% confidence interval [CI], 1.10 to 8.58; P = .03). Moreover, patients with "increased" sL-selectin levels (>/=3.12 microgram/mL) had shorter event-free survival (EFS) (median 7.3 v 12 months, P = .008) and overall survival (median 1 v 2.05 years, P = .03) than patients with sL-selectin <3.12 microgram/mL. Multivariate statistical analysis (adjusted for age and presence of Auer rods) indicated that sL-selectin was an independent prognostic factor for EFS (hazard ratio [HR], 1.96; 95% CI, 1.21 to 3.17, P = .006) and overall survival (HR, 1.80; 95% CI, 1.09 to 2.98; P = .02). Thus, plasma sL-selectin may be a useful prognostic marker in the evaluation of AML at diagnosis.