A robust socioeconomic gradient in health is well-documented, with higher socioeconomic status (SES) associated with better health across the SES spectrum. However, recent studies of U.S. racial/ethnic minorities and immigrants show complex SES-health patterns (e.g., flat gradients), with individuals of low SES having similar or better health than their richer, U.S.-born and more acculturated counterparts, a so-called "epidemiological paradox" or "immigrant health paradox". To examine whether this exists among Asian Americans, we investigate how nativity and occupational class (white-collar, blue-collar, service, unemployed) are associated with subjective health (self-rated physical health, self-rated mental health) and 12-month DSM-IV mental disorders (any mental disorder, anxiety, depression). We analyzed data from 1530 Asian respondents to the 2002-2003 National Latino and Asian American Study in the labor force using hierarchical multivariate logistic regression models controlling for confounders, subjective social status (SSS), material and psychosocial factors theorized to explain health inequalities. Compared to U.S.-born Asians, immigrants had worse socioeconomic profiles, and controlling for age and gender, increased odds for reporting fair/poor mental health and decreased odds for any DSM-IV mental disorder and anxiety. No strong occupational class-health gradients were found. The foreign-born health-protective effect persisted after controlling for SSS but became nonsignificant after controlling for material and psychosocial factors. Speaking fair/poor English was strongly associated with all outcomes. Material and psychosocial factors were associated with some outcomes - perceived financial need with subjective health, uninsurance with self-rated mental health and depression, social support, discrimination and acculturative stress with all or most DSM-IV outcomes. Our findings caution against using terms like "immigrant health paradox" which oversimplify complex patterns and mask negative outcomes among underserved sub-groups (e.g., speaking fair/poor English, experiencing acculturative stress). We discuss implications for better measurement of SES and health given the absence of a gradient and seemingly contradictory finding of nativity-related differences in self-rated health and DSM-IV mental disorders.