Abstract Background In the general population, an increase in low-density lipoprotein cholesterol (LDL-C) predicts higher cardiovascular disease risk, and lowering LDL-C can prevent cardiovascular disease and reduces mortality risk. Interestingly, in cohort studies that include very old populations, no or inverse associations between LDL-C and mortality have been observed. This study aims to investigate whether the association between LDL-C and mortality in the very old is modified by a composite fitness score. Methods A 2-stage meta-analysis of individual participant data from the 5 observational cohort studies. The composite fitness score was operationalized by performance on a combination of 4 markers: functional ability, cognitive function, grip strength, and morbidity. We pooled hazard ratios (HR) from Cox proportional-hazards models for 5-year mortality risk for a 1 mmol/L increase in LDL-C. Models were stratified by high/low composite fitness score. Results Composite fitness scores were calculated for 2 317 participants (median 85 years, 60% females participants), of which 994 (42.9%) had a high composite fitness score, and 694 (30.0%) had a low-composite fitness score. There was an inverse association between LDL-C and 5-year mortality risk (HR 0.87 95% CI: 0.80–0.94; p < .01), most pronounced in participants with a low-composite fitness score (HR 0.85 95% CI: 0.75–0.96; p = .01), compared to those with a high composite fitness score (HR = 0.98 95% CI: 0.83–1.15; p = .78), the test for subgroups differences was not significant. Conclusions In this very old population, there was an inverse association between LDL-C and all-cause mortality, which was most pronounced in participants with a low-composite fitness scores.