Frailty is a well-established risk factor for adverse health outcomes. However, comparatively little is known about the dynamic nature of frailty and the extent to which it can improve. The purposes of this study were to systematically search for studies examining frailty transitions over time among community-dwelling older people, and to synthesise pooled frailty transitions rates. Four electronic databases (Medline, Embase, PsycINFO and CINAHL) were searched in July 2018. Inclusion criteria were: prospective design, community-dwelling older people with mean age>60, using 5-item frailty phenotype criteria to define three states: robust, prefrail and frail and the numbers of participants with 9 frailty transition patterns based on frailty status at baseline and follow-up. Exclusion criteria were: selected populations, using fewer than 5 frailty phenotype criteria. Two investigators independently screened 504 studies for eligibility and identified 16 studies for this review. Data were extracted by the two investigators independently. Pooled rates of frailty transition patterns were calculated by random-effects meta-analysis. Among 42,775 community-dwelling older people from 16 studies with a mean follow-up of 3.9 years (range: 1-10 years), 13.7% (95%CI = 11.7-15.8%) improved, 29.1% (95%CI = 25.9-32.5%) worsened and 56.5% (95%CI = 54.2-58.8%) maintained the same frailty status. Among those who were robust at baseline, pooled rates of remaining robust or transitioning to prefrail and frail were 54.0% (95%CI = 48.8-59.1%), 40.6% (95%CI = 36.7-44.7%) and 4.5% (95%CI = 3.2-6.1%), respectively. Among those who were prefrail at baseline, corresponding rates to robust, prefrail and frail were 23.1% (95%CI = 18.8-27.6%), 58.2% (95%CI = 55.6-60.7%) and 18.2% (95%CI = 14.9-21.7%), respectively. Among those who were frail at baseline, pooled rates of transitioning to robust, prefrail and remaining frail were 3.3% (95%CI = 1.6-5.5%), 40.3% (95%CI = 34.6-46.1%) and 54.5% (95%CI = 47.6-61.3%), respectively. Stratified and meta-regression analyses showed age, gender and follow-up period were associated with frailty transition patterns. Older people make dynamic changes in their frailty status. Given that while one quarter of prefrail older people improved to robust only 3% of frail older people did, early interventions should be considered.