Objectives: Relative and absolute reliability are psychometric properties of the test that many clinical decisions are based on them. In many cases, only relative reliability takes into consideration while the absolute reliability is also very important.

Methods & Materials: Eleven community-dwelling older adults aged 65 years and older (69.64±3.58) and 20 healthy young in the age range 20 to 35 years (28.80±4.15) using three versions of Timed Up and Go test were evaluated twice with an interval of 2 to 5 days.

Results: Generally, the non-homogeneity of the study population was stratified to increase the Intra-class Correlation Coefficient (ICC) this coefficient in elderly people is greater than young people and with a secondary task is reduced. In This study, absolute reliability indices using different data sources and equations lead to in more or less similar results. At general, in test–retest situations, the elderly more than the young people must be changed to be interpreted as a real change, not random. The random error contribution is slightly greater in elderly than young and with a secondary task is increased.It seems, heterogeneity leads to moderation in absolute reliability indices.

Conclusion: In relative reliability studies, researchers and clinicians should pay attention to factors such as homogeneity of population and etc. As well as, absolute reliability beside relative reliability is needed and necessary in clinical decision making.