BACKGROUND: Testosterone, cortisol and their ratios may be indicators of anabolic status, but technical issues surrounding blood sampling has limited wider application. The advent of salivary testosterone (sal-T) analysis simplified sample acquisition, resulting in a subsequent rapid increase in the number of published research articles. OBJECTIVE: The objective of this study was to undertake a meta-analysis to determine the effect of acute exercise bouts on post exercise sal-T and salivary cortisol (sal-C) concentrations and their ratio (sal-T:C). DATA SOURCES: Relevant databases such as PubMed, Web of Science, Science Direct and SPORTDiscus were searched up to and including 31 December 2013 for the term 'saliva AND testosterone AND exercise'. STUDY SELECTION: Studies (n = 21) selected from the 933 identified included randomised controlled trials (RCTs; n = 2), uncontrolled trials (UCTs; n = 18) and control trials (CTs; n = 1), all of which had an exercise component characterised as either aerobic, resistance or power training, each with acute sal-T and sal-C measurement obtained within 30 min of exercise bout completion. STUDY APPRAISAL AND SYNTHESIS METHODS: A meta-analysis was conducted on change in sal-T, sal-C and the sal-T:C ratio following exercise using standard difference in means (SDM) and a random effects model. RESULTS: For aerobic, resistance and power exercise, the overall SDMs for sal-T were 0.891, 1.061 and 0.509, respectively; for sal-C, the SDMs were 3.041, 0.773 and 1.200, respectively. For sal-T:C, the SDMs were -2.014, 0.027 and -0.968, respectively. RCTs, UCTs and CTs were separated by subgroup analysis. There were significant differences in overall weighted SDM values for sal-T between RCTs, UCTs and CTs within exercise modes. When examining aerobic exercise interventions, a quantitative interaction of study design was observed. RCTs resulted in a greater SDM than UCTs (1.337 vs. 0.446). Power interventions displayed a qualitative interaction with study design. UCTs where baseline measures were obtained 24 h before exercise had an SDM of -1.128, whereas UCTs where baseline was determined immediately prior to exercise had an SDM of 0.486. The single CT trial had an SDM of 2.260. Resistance exercise interventions were primarily UCTs; however, an observed influence of baseline sampling time whereby immediately pre- and 24 h pre-exercise resulted in differing SDMs. The sole resistance exercise RCTs resulted in the greatest SDM (2.500). CONCLUSION: The current body of evidence regarding acute responses of sal-T to exercise is weak. This meta-analysis identifies varying exercise-dependent effect sizes. Each appear to be greatly influenced by study design and sample timing. There is a need for more RCTs and a standardised methodology for the measurement of salivary hormones in order to better determine the effect of exercise modality.