Objective: To determine the optimal duration of a run-in period for initiation of real-time continuous glucose monitoring (CGM) before the start of a randomized controlled trial (RCT) in type 1 diabetes (T1D) or type 2 diabetes (T2D). Methods: Data sets were pooled from 8 RCTs, which had a blinded CGM wear period followed by at least 3 months of unblinded CGM use. Across all participants, mean time in range 70-180 mg/dL (TIR) and mean time <54 mg/dL (T < 54) as well as other key CGM metrics were computed for the initial period of blinded CGM wear and from the subsequent 13 weeks of unblinded CGM use. Results: The analysis cohort included data from 485 participants: 348 with T1D and 137 with T2D, ranging in age from 2 to 82 years. Mean TIR was 49% with blinded CGM before initiation of unblinded CGM use, increased to 55% by the end of the first week of unblinded CGM use, and then showed little change through 13 weeks. Mean T < 54 decreased from 1.4% with blinded CGM to 0.8% 1 week and 0.6% 2 weeks after initiating unblinded CGM use, which matched the value in month 3. Similar results were obtained for mean glucose, time >180 mg/dL, time >250 mg/dL, and time <70 mg/dL, with the mean improvement in hyperglycemia metrics plateauing slightly faster than hypoglycemia metrics. Findings were largely similar for T1D and T2D. Conclusion: When initiating unblinded real-time CGM, improvement in key CGM metrics occurs rapidly, with maximal effect on the mean of each metric achieved within 1-2 weeks. For a randomized trial in which all participants will use real-time unblinded CGM for glucose monitoring, a run-in period should be implemented before collecting baseline data for participants who are not CGM users. For such CGM-naive individuals, a 7- to 14-day acclimation period is sufficient followed by a 14-day period for collection of baseline unblinded CGM data.