BACKGROUND: In the intensive care unit (ICU), few data are available to guide the maximum infusion rate for individual vasopressors or vasopressor combinations. Clinician preference, decisions to use multiple vasopressors, safety concerns and institutional policy may all affect maximal vasopressor dosing thresholds and potentially contribute to variation in clinical practice. We sought to describe clinical practice patterns in the maximum dosing of vasopressors used as part of the management of patients with shock. We expected to find variability between ICUs in maximum infusion rates for both individual vasopressors and the cumulative dose of vasopressor combinations. METHODS: Retrospective cohort study examining variations in maximum vasopressor doses among adult patients with shock, defined by receipt of one or more vasopressors for at least 6 hours, admitted to 21 ICUs in Alberta, Canada. We defined the maximum vasopressor dose as the highest infusion rate for ≥30 minutes for each individual vasopressor given to a patient, and the maximum total vasopressor dose as the highest combined dose for all vasopressors concurrently administered to a given patient (reported as norepinephrine-equivalents [NEQ]). For each individual vasopressor and overall, we defined a "high-dose" maximum as receipt of a dose above the 90th percentile of maximum doses across the entire patient cohort. We then applied these 90th percentile cutoffs to patients in each ICU. We used multi-level logistic regression to calculate the adjusted median odds ratio (aMOR) for receipt of high-dose maximum NEQ, as a measure of variation of maximum dosing between ICUs. RESULTS: Among 28,397 ICU admissions, the 90th percentile of unadjusted maximum NEQ dose reported administered to patients with shock was 0.7 µg/kg/min (full range 0.01-7.3 µg/kg/min; 1st to 99th percentile 0.03-2.0 µg/kg/min; 10th percentile 0.1 µg/kg/min). The percent of patients in each ICU receiving a high-dose maximum vasopressor (≥0.7 µg/kg/min) ranged from 3.6% to 32.1%. After adjustment for known patient factors, the aMOR (95% confidence interval [CI]) for receipt of a high-dose maximum vasopressors between different ICUs remained: 1.89 (1.50-2.40). Across the entire cohort, the 90th percentile cutoff for high-dose maximum norepinephrine was 0.5 µg/kg/min, for epinephrine 0.5 µg/kg/min, for phenylephrine 316.5 µg/min, for dopamine 15.0 µg/kg/min, and for vasopressin 3.6 units/h. Hospital survival was 31.6% for those who received high-dose vasopressors versus 79.7% for those who received lower doses ( P < .001). We found no clear dosing cutoff above which survival to hospital discharge was impossible. CONCLUSIONS: In Alberta CA, the maximum combined vasopressor dose for patients with shock varied between ICUs after adjustment for severity of illness. Some survival occurred even with high maximal doses. Further work is needed to understand the best approach to determining maximum doses for individual patients at high risk of death.