Critical care for advanced lung cancer patients is still controversial, and the appropriate method for the selection of patients who may benefit from intensive care unit (ICU) care is not clearly defined. We retrospectively reviewed the medical records of stage IIIB-IV lung cancer patients admitted to the medical ICU of a university hospital in Korea between 2003 and 2011. Of 95 patients, 64 (67%) had Eastern Cooperative Oncology Group (ECOG) performance status (PS)≥2, and 79 (84%) had non-small-cell lung cancer. In total, 28 patients (30%) were newly diagnosed or were receiving first-line treatment, and 22 (23%) were refractory or bedridden. Mechanical ventilation was required in 85 patients (90%), and ICU mortality and hospital mortality were 57 and 78%, respectively. According to a multivariate analysis, a PaO2/FiO2 ratio<150 [odds ratio (OR)=5.51, 95% confidence interval (CI) 2.10-14.48, p=0.001] was independently associated with ICU mortality, and an ECOG PS≥2 (OR=9.53, 95% CI 2.03-44.85, p=0.004) and a need for vasoactive agents (OR=6.94, 95% CI 1.61-29.84, p=0.009) were independently associated with hospital mortality. Refractory or bedridden patients (n=22) showed significantly poorer overall survival (11.0 vs. 29.0 days, p=0.005). Among 21 patients who were discharged from the hospital, 11 (52%) received further chemotherapy. Certain advanced lung cancer patients may benefit from ICU management. However, refractory patients and patients with a poor PS do not seem to benefit from ICU care. Oncologists should try to discuss palliative care and end-of-life issues in advance to avoid futile care.