INTRODUCTION: Prolonged intensive care unit lengths of stay (ICU LOS) for critical illness can have acceptable mortality rates and quality of life despite significant costs. Only a few studies have specifically addressed prolonged ICU LOS after trauma. Our goals were to examine characteristics and outcomes of trauma patients with LOS >or= 30 days, predictors of prolonged stay and mortality. METHODS: All trauma ICU admissions over a seven-year period in a level 1 trauma center were analyzed. Admission characteristics, pre-existing conditions and acquired complications in the ICU were recorded. Logistic regression was used to identify independent predictors of prolonged LOS and predictors of mortality among those with prolonged LOS after univariate analyses. RESULTS: Of 4920 ICU admissions, 205 (4%) had ICU LOS >30 days. These patients were older and more severely injured. Age and injury severity score (ISS) were associated with prolonged LOS. After logistic regression analysis, sepsis, acute respiratory distress syndrome, and several infectious complications were important independent predictors of prolonged LOS. Within the group with ICU LOS >30 days, predictors of mortality were age, pre-existing renal disease as well as the development of renal failure requiring dialysis. Overall mortality was 12%. CONCLUSIONS: The majority of patients with ICU LOS >or= 30 days will survive their hospitalization. Infectious and pulmonary complications were predictors of prolonged stay. Further efforts targeting prevention of these complications are warranted.