In vivo expansion of haploidentical natural killer (NK) cell infusions with interleukin-2 (IL-2) can induce remission of refractory acute myeloid leukemia, but efficacy may be hampered by concurrent stimulation of host regulatory T cells. To overcome this limitation, we substituted the NK homeostatic factor IL-15 in 2 phase 1/2 trials. Forty-two patients received either intravenous (IV) (NCT01385423) or subcutaneous (SC) (NCT02395822) recombinant human IL-15 (rhIL-15) after lymphodepleting chemotherapy and haploidentical NK cells. Escalating doses of rhIL-15 (0.3-1.0 μg/kg) were given on 12 consecutive days in a phase 1 trial. Of 26 patients, 36% had robust in vivo NK-cell expansion at day 14, and 32% achieved complete remission. Hypothesizing that SC dosing of rhIL-15 would be safer and better tolerated, 16 patients received 10 once per day doses of SC rhIL-15 at 2.0 μg/kg on a phase 2 trial. NK-cell expansion at day 14 was seen in 27% of the patients, and 40% achieved remission. rhIL-15 induced better rates of in vivo NK-cell expansion and remission compared with previous trials with IL-2, but it was associated with previously unreported cytokine release syndrome (CRS) after SC but not IV dosing. CRS was observed in 56% of patients given SC rhIL-15 (with concurrent neurologic toxicity in 5 of 9 patients) and was responsive to steroids and tocilizumab. SC administration was associated with slower pharmacokinetic clearance and higher levels of IL-6 than IV dosing. These novel trials testing the use of IL-15 to potentiate cell therapy suggest that dosing schedules based on pharmacokinetics and pharmacodynamics will preserve the therapeutic benefits of IL-15 and minimize CRS. These trials were registered at www.clinicaltrials.gov as #NCT01385423 and #NCT02395822.