Using longitudinal data from the China Health and Nutrition Survey, we investigate the causal relation between housing conditions (both internal and external) and health among urban adults aged 18+. We find that housing improvement reduces the probability of bad self-reported health by 3.7 percent, with more pronounced impacts among females, older adults, those with lower socioeconomic status (low education and income) and residents of the less developed central and western regions. This beneficial health effect is enhanced by longer treatment periods and consistent across several robustness checks. Housing conditions seemingly operate on health via poor macronutrient intake, physical inactivity, and sleep deprivation.