OBJECTIVE: The objective of this study was to estimate the association between socioeconomic status (SES) and outcome for admissions to intensive care. RESEARCH DESIGN: Retrospective cohort study. SUBJECTS: We studied 51,572 admissions to 99 intensive-care units in England and Wales between 1995 and 2000. MEASURES: The SES of admissions was measured using Carstairs deprivation scores. Outcome was hospital mortality after adjustment for case mix using the APACHE II method. RESULTS: Admissions of lower SES were, on average, younger and less likely to be following surgery. There was evidence of a SES gradient for hospital mortality in admissions after elective surgery after adjusting for case mix (test for trend P <0.001), with higher SES associated with lower mortality. In the least-deprived quintile of SES, the odds ratio for hospital mortality was 0.70 (95% confidence interval, 0.58-0.84) compared with the most deprived quintile. There was no evidence of a SES gradient for hospital mortality in nonsurgical or emergency surgical admissions, and the decision to withdraw active treatment did not differ by SES. CONCLUSIONS: There is a SES gradient for hospital mortality in elective surgical admissions that is not explained by differences in case mix or the withdrawal of active treatment. Further research is required to establish if this finding can be explained by unmeasured differences in health status at admission to an intensive-care unit or differences in care and to establish the potential impact these results may have on interpreting comparative surgical performance data.