Accelerometers are widely used to measure physical activity (PA) levels, offering insights into PA intensity and volume necessary for health benefits. Often expressed in absolute terms (e.g. metabolic equivalents), activity intensity expressed relative to cardiorespiratory may be more suitable for individuals with varying cardiorespiratory fitness and functional status. Limited device-based methods exist to predict relative PA intensity, often requiring laboratory-based exercise testing for individual-specific method calibration. Additionally, the free-living validity of relative PA estimates using individualized cutpoints has yet to be determined. This study aimed to 1) evaluate the validity of relative PA estimates using individualized cutpoints derived from exercise testing and novel random forest methods developed from free-living activities and 2) compare the dose-response association between PA and cardiovascular disease risk factors (CVDRF) when using absolute or relative intensity. Participants (n = 44, age = 40.8 ± 15.9, 52.3% female) completed laboratory-based exercise testing while undergoing metabolic measurement and completed free-living activities for up to three days while wearing a heart rate monitor on the chest, activPAL on the right thigh, and accelerometers at the right hip and both dominant and non-dominant wrists. Compared to activPAL-derived sedentary behavior and heart rate monitor-determined %HRR (criterion relative PA intensity), intensity-specific classification accuracy for MPA and VPA were low regardless of wear location, ranging from 19% to 33%. Although estimates were biased for MPA and VPA, sensitivity analyses demonstrated that biases were lower during locomotion periods compared to non-locomotion periods using individualized cutpoints. Random forest method estimates of SB and LPA were unbiased. MPA and VPA were significantly underestimated using random forest models by ≥59.9% and did not improve after inclusion of individual-specific covariates. Findings from NHANES data indicated that while both absolute and relative intensity scales showed lower odds of CVDRF with higher MVPA participation, the dose-response association differed between the two scales. Minimal MVPA participation (2-6 minutes/day) was associated with protective benefits on relative intensity scales, while ≥50 minutes/day on absolute intensity scales was required to observe significant risk reduction. Future research should focus on refining methods for accurately assessing relative PA intensity and elucidating its implications for health outcomes.