Tax exemptions for employer-provided health insurance have topped $1 trillion over the past decade, and tax strategies remain a central focus of proposed policies to expand health insurance coverage. In this paper I estimate the impact of the tax-preferred treatment of health insurance on the level of health insurance coverage. I employ a pair of restricted government datasets comprising the Agency on Health Research and Quality's Medical Expenditure Panel Survey. These data, unique in their content, allow me to address one of the primary challenges in this field. There has been difficulty in identifying the numerous ways in which both firms and households may respond to the tax incentives. These include whether insurance is offered by the employer, whether firms change plan details, and whether households participate in offered benefits. In addition, the data allow for consideration of important variables, such as the distribution of workers within a firm and individuals' health statuses, which are not available in publicly available data. Examining changes in tax rates at the start of the decade, I find that some firms have responded to decreases in marginal tax rates by not offering benefits. This response is employed by between 1% and 1.5% of firms and is restricted almost exclusively to small firms. In addition, a small percentage of workers have responded by declining coverage. This behavior occurs primarily among younger and low-income workers and is present in all but the smallest firms. In contrast to these two responses, the level of spending per participant has not responded to changes in tax rates. This does not support the concerns of many that the tax treatment of employer-provided health insurance leads to Ì¢åÛåÏover-consumptionÌ¢åÛå of health insurance. In addition to the responses to changes in tax rates, I find that the earnings distribution of workers within a firm is one of the strongest determinants of whether firms offer health insurance, the number of workers that decline coverage and the level of spending per participant. In total, these findings imply that beneficial tax treatment of benefits does raise health insurance coverage, but the size and composition of the risk pool are far more important characteristics.