Previous studies of the characteristics of the population without health insurance have relied almost exclusively on data from a point in time (e.g., Swartz; Wilensky; Chollet; Moyen; Congressional Research Service). Missing from these analyses is information on the distribution of the lengths of time people are without health insurance. To devise policies related to the uninsured, policy makers need to know both who is without health insurance and how long they are without health insurance. If the vast majority of people with long uninsured spells do not have strong attachments to the labor force, then mandating employer provided health insurance is not likely to provide health insurance to great numbers of the uninsured. Similarly, if most of the people with long uninsured spells are poor or near-poor, then an expansion of Medicaid might be a more effective policy for providing financial access to medical care to the uninsured. And, if many people with short uninsured spells have the spells because they are changing jobs, then a policy that provides for continuation of coverage for at least three months past the termination of employment might be effective.