Objective: To study the association between physical fitness and body-size phenotypes, and to test which aspects of physical fitness show the greatest independent association with cardiometabolic risk in perimenopausal women. Study design: This cross-sectional study involved 228 women aged 53 +/- 5 years from southern Spain. Main outcome measurements: Physical fitness was assessed by means of the Senior Fitness Test Battery (additionally including handgrip strength and timed up-and-go tests). Anthropometry, resting heart rate, blood pressure and plasma markers of lipid, glycaemic and inflammatory status were measured by standard procedures. The harmonized definition of the 'metabolically healthy but obese' (MHO) phenotype was employed to classify individuals. Results: The overall prevalence of the MHO phenotype was 13% but was 43% among the obese women. Apart from traditional markers, metabolically healthy non-obese women had lower levels of C-reactive protein than women with the other phenotypes (p <0.001), and levels of glycosylated haemoglobin were lower in MHO women than in metabolically abnormal non-obese women (overall p = 0.004). Most of the components of physical fitness differed with body-size phenotypes. The 6-min walk and the back-scratch tests presented the most robust differences (both p <0.001). Moreover, the women's performance on the back-scratch (beta = 0.32; p <0.001) and the 6-min walk (beta = 0.22; p = 0.003) tests was independently associated with the clustered cardiometabolic risk. The back-scratch test explained 10% of the variability (step 1, p <0.001), and the final model, which also included the 6-min walk test (step 2, p = 0.003), explained 14% of the variability. Conclusion: Low upper-body flexibility was the most important fitness indicator of cardiometabolic risk in perimenopausal women, but cardiorespiratory fitness also played an important role. (C) 2016 Elsevier Ireland Ltd. All rights reserved.