The way poverty is measured shapes the types of policy solutions perceived to be possible and appropriate to address poverty, as poverty measurement produces information about who is poor, how many people are poor, and why they are poor. The traditional approach to measuring poverty in the United States suffers from two serious shortcomings, which limit the usefulness of the data produced to inform poverty policy. First, the official federal poverty measure (OPM) traditionally used to determine who qualifies as poor is based on consumption data from the 1960s and does not reflect current living patterns or costs of basic needs. Second, poverty in the United States is typically measured on an annual basis, using a cross-sectional analysis approach, which fails to capture information about the duration of poverty, though short-term poverty and long-term poverty have been shown to have different demographics, and long-term poverty is associated with more severe impacts on life outcomes. This study addresses these two shortcomings, by using an alternative poverty measure recently developed by the U.S. Census Bureau and U.S. Bureau of Labor Statistics, the Supplemental Poverty Measure (SPM), in place of the OPM to determine who qualifies as poor, and by analyzing poverty from a longitudinal rather than cross-sectional perspective, examining chronic or long-term poverty and transient or short-term poverty as distinct phenomena. Prior research has examined poverty in the U.S. using alternative poverty measures including the SPM, but only from a cross-sectional perspective. Other research has examined U.S. poverty from a longitudinal perspective, but using the OPM or a closely derived poverty measure. This study thus fills a gap in the existing research on poverty in the United States, by measuring poverty longitudinally using the better-grounded Supplemental Poverty Measure. Data for this study were drawn from the Panel Study of Income Dynamics (PSID), a comprehensive nationally representative longitudinal dataset. Data included detailed household income, benefit, housing, and expense information used to construct annual poverty status using the SPM, as well as individual and household demographic variables, collected biennially from 1998 to 2008, thus representing six data years. Descriptive analysis was conducted using individuals as the unit of analysis (n= 8,375) while multivariate regression analysis was conducted using households as the unit of analysis (n=4,188). Complex survey weights were used in all analyses to adjust for differential sampling and attrition. Multiple imputation was used to impute missing values for one of the components used to construct SPM poverty status and for one of the demographic covariates. Chronic poverty was defined as poor under the SPM in more than half of the years examined (i.e. 4 or more of 6 years), while transient poverty was defined as poor under the SPM in at least one year but not more than half of the years (i.e. 1 to 3 of 6 years). Nonpoor was defined as not poor under the SPM in any year. Descriptive analysis was used to examine the prevalence and demographics of chronic and transient poverty, to compare the demographics of chronic and transient poverty using the Supplemental Poverty Measure versus using the official federal poverty measure, and to examine the impact of existing government benefits, private resources, and household expenses on chronic and transient poverty rates. Results showed that chronic poverty was a rare phenomenon, affecting only 2.1% of the sample or approximately 1 in 50 individuals, while transient poverty was fairly common, affecting 18.9% of the sample or approximately 1 in 20 individuals. The demographics of chronic and transient poverty were somewhat different, with groups that experienced high rates of transient poverty generally demonstrating even more disproportionately high rates of chronic poverty. Thus chronic poverty was more concentrated among particularly disadvantaged groups, while the population affected by transient poverty was still disadvantaged but more similar to the overall sample. The rates of chronic and transient poverty calculated using the SPM were statistically significantly different from the rates calculated using the official federal poverty measure, for both the overall sample and for many demographic subgroups. In general, chronic poverty rates were lower, and transient poverty rates were higher, when using the SPM versus using the OPM. Finally, government benefits were shown to have a substantial impact on both chronic and transient poverty rates, reducing the overall transient poverty rate from 23.9% to 18.9%, a difference of 5.0 percentage points, and reducing the overall chronic poverty rate from 10.8% to 2.1%, a reduction of 8.7 percentage points. One observed effect of government benefits was to increase household resources just enough to shift some individuals out of chronic poverty into transient poverty. The impact of government benefits on chronic and transient poverty rates was different for different demographic subgroups. Seniors experienced the greatest reduction in transient and especially chronic poverty rates, essentially due to Social Security, while children experienced less of a reduction. For immigrants, the dominant effect of government benefits was to shift individuals out of chronic into transient poverty. Multivariate regression, specifically multinomial logistic regression, was used to examine the predictors of transient and chronic poverty. Analysis specifically examined whether the predictors of each type of poverty, versus nonpoor status, corresponded to economic theory which posits that transient poverty is driven by temporary reductions in income (e.g. job layoff), while chronic poverty is driven by an inadequate long-term base of human and material assets needed to generate income (e.g. lack of education or presence of disability). Results showed that chronic poverty was significantly associated with asset limitations, including particularly non-high school graduate status, immigrant status, and long-term disability in a high housing cost area. Transient poverty was significantly associated with one variable linked to short-term income disruption, namely short-term unemployment. However, transient poverty was also significantly predicted by variables representing asset limitations, though most of these covariates had a stronger association with chronic poverty than transient poverty. The association of asset limitations with transient poverty appeared to be partly explained by the fact that government benefits shifted some asset-limited households, who would be expected to be chronically poor, out of chronic poverty and into transient poverty. Results of this study suggest implications for both research and policy. The finding that rates of chronic and transient poverty differ depending on whether the Supplemental Poverty Measure or official federal poverty measure is used suggests that researchers and policy analysts should consider using the SPM when analyzing longitudinal poverty, as the SPM has a stronger conceptual and empirical grounding than the OPM and did not simply function as a proxy for the OPM when examining poverty longitudinally in this study. Results related to the impact of government benefits on chronic and transient poverty rates suggest that policymakers should consider not just short-term policy impacts, but also the longitudinal impact of specific policies and of the overall package of government benefits on poverty. In addition, the differential impact of policies on chronic versus transient poverty, and on chronic and transient poverty among different demographic subgroups, should be considered. Findings related to the predictors of chronic versus transient poverty suggest that policies to address chronic poverty should target individuals with limited bases of human assets needed to generate income; such policies could function either through asset building or through long-term income supplementation or subsidies. Transient poverty could be addressed by enhancing short-term unemployment support, while policies targeted to asset-limited individuals would be likely to impact transient as well as chronic poverty. Further research to more clearly distinguish predictors of chronic poverty over and above transient poverty would be helpful for policy targeting purposes. Finally, prior research on the impact of chronic and transient poverty on life outcomes suggests that two types of poverty could be considered as priorities for policy interventions, due to greater impact on health and other outcomes, namely chronic poverty (as exposure to longer duration of poverty is associated with worse outcomes) and transient poverty occurring during the sensitive developmental period of childhood (as exposure to even short-term poverty during this sensitive period is associated with serious long-term health and developmental impacts). Results from this study show that addressing either of these two types of poverty could be feasible, if somewhat ambitious policy goals in terms of the number of individuals affected and the cumulative gap between their resources and needs.