Healthcare performance assessment is especially relevant for Intensive Care Units (ICUs), which deal with high complexity cases. This work evaluates 93 ICUs using Data Envelopment Analysis (DEA). Three models are proposed to broaden the analysis from different perspectives: staffing, structure, and capacity. It uses patient-level data to adjust outcomes to the ICU s case-mix, which results in two outputs: Standardized Mortality Rate (SMR) and Standardized Resource Use (SRU). Statistical analyses are also performed to show the relation of the variables. Average DEA efficiency scores are calculated for categorical non-discretionary variables, assessing that private for-profit hospitals, in general, have better efficiency results and that large ICUs have lower SMR and SRU.