Background: Treadmills are routinely used to assess running performance and training parameters related to physiological or perceived effort. These measurements are presumed to replicate overground running but there has been no systematic review comparing performance, physiology and perceived effort between treadmill and overground running. Objective: The objective of this systematic review was to compare physiological, perceptual and performance measures between treadmill and overground running in healthy adults. Methods: AMED (Allied and Contemporary Medicine), CINAHL (Cumulative Index to Nursing and Allied Health), EMBASE, MEDLINE, SCOPUS, SPORTDiscus and Web of Science databases were searched from inception until May 2018. Included studies used a crossover study design to compare physiological (oxygen uptake [V˙O2 ], heart rate [HR], blood lactate concentration [La]), perceptual (rating of perceived exertion [RPE] and preferred speed) or running endurance and sprint performance (i.e. time trial duration or sprint speed) outcomes between treadmill (motorised or non-motorised) and overground running. Physiological outcomes were considered across submaximal, near-maximal and maximal running intensity subgroups. Meta-analyses were used to determine mean difference (MD) or standardised MD (SMD) ± 95% confidence intervals. Results: Thirty-four studies were included. Twelve studies used a 1% grade for the treadmill condition and three used grades > 1%. Similar V˙O2 but lower La occurred during submaximal motorised treadmill running at 0% (V˙O2 MD: – 0.55 ± 0.93 mL/kg/min; La MD: − 1.26 ± 0.71 mmol/L) and 1% (V˙O2 MD: 0.37 ± 1.12 mL/kg/min; La MD: − 0.52 ± 0.50 mmol/L) grade than during overground running. HR and RPE during motorised treadmill running were higher at faster submaximal speeds and lower at slower submaximal speeds than during overground running. V˙O2 (MD: − 1.25 ± 2.09 mL/kg/min) and La (MD: − 0.54 ± 0.63 mmol/L) tended to be lower, but HR (MD: 0 ± 1 bpm), and RPE (MD: – 0.4 ± 2.0 units [6–20 scale]) were similar during near-maximal motorised treadmill running to during overground running. Maximal motorised treadmill running caused similar V˙O2 (MD: 0.78 ± 1.55 mL/kg/min) and HR (MD: − 1 ± 2 bpm) to overground running. Endurance performance was poorer (SMD: − 0.50 ± 0.36) on a motorised treadmill than overground but sprint performance varied considerably and was not significantly different (MD: − 1.4 ± 5.8 km/h). Conclusions: Some, but not all, variables differ between treadmill and overground running, and may be dependent on the running speed at which they are assessed. Protocol registration: CRD42017074640 (PROSPERO International Prospective Register of Systematic Reviews).