OBJECTIVES: Recent studies have documented high rates of non-administration of ordered venous thromboembolism (VTE) prophylaxis doses. Intervention strategies that target all patients have been effective, but prohibitively resource-intensive. We aimed to identify efficient intervention strategies based on patterns of non-administration of ordered VTE prophylaxis. METHODS: In this retrospective review of electronic medication administration records, we included adult hospitalized patients who were ordered pharmacologic VTE prophylaxis with unfractionated heparin or enoxaparin over a seven-month period. The primary measure was the proportion of ordered doses of VTE prophylaxis not administered, assessed at the patient, floor, and floor type levels.  Differences in non-administration rates between groups were assessed using generalized estimating equations. RESULTS: A total of 103,160 ordered VTE prophylaxis doses during 10,516 patient visits on twenty-nine patient floors were analyzed.  Overall, 11.9% of ordered doses were not administered.  Approximately 19% of patients missed at least one quarter and 8% of patients missed over one half of ordered doses.  There was marked heterogeneity in non-administration rate at the floor level (range: 5-27%).  Patients on medicine floors missed a significantly larger proportion (18%) of ordered doses compared to patients on other floor types (8%, Odds Ratio: 2.4, p<0.0001). However, more than half of patients received at least 86% of their ordered doses, even on the lowest performing floor.  The 20% of patients who missed at least two ordered doses accounted for 80% of all missed doses. CONCLUSIONS: A substantial proportion of ordered doses of VTE prophylaxis were not administered.  The heterogeneity in non-administration rate between patients, floors, and floor types can be used to target interventions.  The small proportion of patients that missed multiple ordered doses accounted for a large majority of non-administered doses.  This recognition of the Pareto principle provides opportunity to efficiently target a relatively small group of patients for intervention.