A fundamental challenge for labor-intensive firms facing demand uncertainty is how to configure their workforce to improve quality and performance. This dissertation investigates determinants of firms' workforce configuration, with a focus on the mix of temporary and permanent nurses, and its implications on patient outcomes, a central measure of performance in health care. This is a timely topic since many industries including health care are increasingly using alternative work arrangements. This dissertation uses novel and rich proprietary data from a large US freestanding home health company. The data provide detailed information on both the consumption and production sides of home health care delivery: utilization, referral sources, risk factors and health outcomes of patients as well as nurses’ work logs and human resources characteristics. In Chapter 1, I investigate the effect of firms’ labor mix on patient readmission, a main quality marker in post-acute care, using exogenous variation in full-time nurses’ activeness in the patient’s neighborhood and unavailability of nearest full-time nurses. I find that patients who received one standard-deviation higher proportion of full-time nurse visits were 7 percent less likely to be readmitted. In Chapter 2, I investigate the effect of reputation on firms’ labor mix strategy under demand uncertainty. Firms face a trade-off in using temporary nurses: they provide flexibility in responding to demand fluctuations but may impede the establishment of reputation through lower quality of service. I present and test a model of firm’s labor mix choices where the labor mix dynamically shapes the market’s perception of the firm’s quality (i.e. reputation), and the firm’s demand is in turn stochastically linked to its reputation. I find evidence consistent with the model: firms with lower reputation, especially young firms, increased the percentage of full-time nurses with demand volatility. Young firms trade off short-term profitability for long-term reputation gains. In Chapter 3, I investigate the effect of firms’ workforce assignment on readmission through care discontinuity or handoffs. Using exogenous variation in workflow interruption caused by providers’ inactivity, I find handoffs to increase the likelihood of readmission. One in four readmissions during home health would be avoided if handoffs were eliminated.