This dissertation contains three capacity management problems in health care delivery systems. In particular, Chapter 2 evaluates a panel design problem regarding how clinics may wish to best allocate a pool of heterogeneous patients (i.e. non-acute and acute patients) into physician panels. The analytical results show that neither specialization (i.e. each panel contains patients that are as homogeneous as possible) or equal assignment (i.e. identical panels with same types of patient mix) is a dominant patient allocation strategy. The results also show that equal assignment strategy works better when acute demand is relatively low or high as compared to the capacity, and specialization works better when acute demand is moderate. This chapter serves to highlight the impact of patient composition on the performance of a clinic profile. Chapter 3 investigates how clinics may learn and utilize patients' preference information through an existing web-based interface in appointment booking decisions. Analytical results leading to a partial characterization of an optimal booking policy are presented. Examples show that heuristic decision rules, based on this characterization, perform well and reveal insights about trade-offs among a variety of performance metrics such as expected revenue, patient-PCP match rate, number of patients served, and capacity spoilage rate. Chapter 4 focuses on identifying observable predictors of nurse absenteeism and incorporates these factors into staffing decisions. The analysis highlights the importance of paying attention to unit-level factors and absentee-rate heterogeneity among individual nurses. The data-based investigation confirms that nurses' absence history is a good predictor of their future absences. This result is used as the nurse absenteeism assumption in the model-based investigation that evaluates how to assign nurses to identical nursing units when nurses' absentee rates are heterogeneous. We propose and test several easy-to-use heuristics to identify near optimal staffing strategies for inpatient units.