Uncertainty in treatment durations is a critical factor in chemotherapy appointment scheduling (CAS). We first examine the sources of variability in treatment times and illustrate their scope using real data from an Outpatient Chemotherapy Unit (OCU). Building on this motivation, we present several two-stage stochastic mixed-integer programming (SMIP) formulations adapted from the literature. We begin with a base model that balances expected patient waiting time and nurse overtime. Recognizing its limitations, we then introduce successive extensions that incorporate daily nurse workload balancing, manage instantaneous nurse workload through acuity-based measures, allow for varying levels of nursing care flexibility, and coordinate oncologist consultations with chemotherapy appointments.