Objective: To examine the short and long-term effects of exercise on glucose outcomes as assessed by continuous glucose monitors (CGM) in type 2 diabetes (T2D). Methods: A search of databases was performed up to May 2018. Eligible studies had participants complete standardized exercise interventions using CGM. In all short-term studies (<2 weeks, often a single exercise session) participants completed both an exercise and non-exercise control condition. In long-term studies (>8 weeks) the difference pre and post exercise training values were compared. The within participant mean differences and standard errors in 24-hour glucose concentrations between conditions was used to obtain a weighted mean difference according to the Generic Inverse Variance method (RevMan version 5.3). The I statistic was calculated to represent the percentage of the variability due to heterogeneity. Results: A total of 27 studies were included. Of these, 22 studies were short-term, and 5 studies were longterm interventions. Exercise decreased 24-hour glucose in short-term (-0.6 mmol/L; -0.8 to -0.4; p<0.01, I =79%) and long-term studies (-0.5 mmol/L; -0.7 to -0.2; p<0.01, I = 0%). To explain the heterogeneity in the short-term studies, subgroup and meta-regression analyses were performed. Comparison of subgroups defined according to the timing of exercise in relation to meals revealed significant differences among subgroups. In the 11 studies that performed exercise after breakfast, 24-hour mean glucose was reduced by 0.7 mmol/L (-0.9 to -0.5). In the 3 studies that performed exercise before dinner, there was no effect on 24- hour glucose (0.0 mmol/L -0.2 to 0.2). Elevated glucose in the control condition predicted a greater decrease in mean 24-hour glucose following exercise (R = 0.29, P=0.0001). Conclusion: Both short and long-term exercise can lower glucose in individuals with T2D. The timing of exercise as well as non-exercise glucose values were predictors of glycemic response to short-term exercise.