The assessment of weight loss as an indicator of poor nutritional status in older persons is currently widely applied to establish risk of mortality. Little is known about the relationship between changes in mid-upper arm circumference (MUAC) and mortality in older individuals. The aim of the present study was to examine the association between 3-year change in MUAC and 20-year mortality in community-dwelling older adults and compare this to the association between body mass index (BMI) change and mortality. Data on changes in MUAC (cm) and BMI (kg/m2), covariates, and mortality were available for 1307 Dutch older adults (49.7% men) aged 65 years and older in 1995/96 (mean 75.6 years, SD 6.5) from Longitudinal Aging Study Amsterdam (LASA). Anthropometric measurements were performed in 1992/93 with repeated measurements in 1995/96 (baseline), and a mortality follow up until July 2015. BMI and MUAC change were divided into quintiles, with the quintile including zero defined as the reference category. Cox regression analyses were performed to examine the associations of 3-year changes in MUAC and BMI with subsequent 20-year all-cause mortality, adjusted for demographic and health factors. Age, sex and initial measurement of BMI and MUAC (1992/93) were tested for effect modification (P=<0.10). Mean baseline BMI was 26.7kg/m2 (SD 4.2) with a 3-year change of-0.2 (SD 1.5). Mean baseline MUAC was 30.5cm (SD 3.5) with a 3-year change of-0.8 (SD 1.6). Age, sex, and BMI and MUAC 3 years prior were effect modifiers in the associations between change in anthropometric measurement and mortality. Decrease in MUAC was not associated with mortality in persons with a higher initial MUAC (≥31cm), while for persons with a lower initial MUAC, a decrease in MUAC of ≤-2.15 was associated with increased mortality risk (HR 1.54; 95% CI: 1.14-2.09), also when further stratified on median age and sex. In stratified analysis of BMI change for median initial BMI (26.5) and additionally stratified for median age and sex, the associations between a BMI decrease of ≤-1.19 and mortality fluctuated, mostly statistically not significant. No associations were found for gain in MUAC or BMI. Given that MUAC loss is more strongly and consistently associated with an increased mortality risk in older individuals with a low initial MUAC compared to BMI loss, this may be a more recommendable measure to use in clinical practice for assessing poor nutritional status, instead of weight loss.