Abstract: This is a survey study that aims to determine the effect of expectations and social support on well-being. The population is poor household (Keluarga Tidak Mampu/KTM) which was recorded in 2010 in the office of community development (Kantor Pemberdayaan Masyarakat/KPM) Bontang, East Kalimantan. From this study population are 8348 poor households (KTM), a sample taken 1500 poor households representing 18% of the total population of this study. Methodologically, the study sample size was considered to represent the population of existing research. The sampling technique research done by cluster random sampling based on the amount of data in each village poor household (KTM) available, samples were taken at random by the same percentage amount, namely 18%. The sample of this research is focused on the head of each poor household with 10 variations in the type of research, namely, the driver, fishermen, traders, construction workers, farmers, ranchers, workers, motorcycle taxi drivers, night-watchman and domestic servants. Determination of the 10 variations of this type of research is based on data obtained from the office of community development (Kantor Pemberdayaan Masyarakat/KPM) Bontang to successfully identify the types of jobs from poor household (KTM) recorded in 2010. Methods of data collection in this study was conducted using self-report methods (self-report) through a questionnaire with open and closed answers. Measurement instruments used in this research questionnaire contains 3 scales of measurement, namely, the scale of welfare Riff, the scale of expectation and social support scale which is based on the framework of house and kahn. The results showed a correlation coefficient between variables life expectancy (hope) with the welfare (SWB) of 0.678 while the social support (Duksos) variables with the welfare (SWB) of 0.513. The results showed the effect of variable levels of expectations (hope) and social support (Duksos) on well-being in which the results of multiple regression analysis showed F-value of 606 271 with p = 0.000 (p<0.05). The coefficient of determination of this study show number 0.466 of which 46.6% welfare (SWB) variability can be explained by the variable level of expectations (hope) and social support (Duksos). Standard error of estimate model of this study indicate the numbers 5.842 where the rate is smaller than the standard deviation of the welfare (SWB) variable (7.987), so that, the regression model can actually function as a predictor variable welfare (SWB).