We used data from a longitudinal observation study to determine whether markers of cartilage turnover could serve as predictors of cartilage loss on magnetic resonance imaging (MRI). We conducted a study of data from the Boston Osteoarthritis of the Knee Study (BOKS), a completed natural history study of knee osteoarthritis (OA). All subjects in the study met American College of Rheumatology criteria for knee OA. Baseline and follow-up knee magnetic resonance images were scored for cartilage loss by means of the WORMS (Whole Organ Magnetic Resonance Imaging Score) semiquantitative grading scheme. Within the BOKS population, 80 subjects who experienced cartilage loss and 80 subjects who did not were selected for the purposes of this nested case control study. We assessed the baseline levels of cartilage degradation and synthesis products by means of assays for type I and II cleavage by collagenases (Col2:3/4C(short) or C1,2C), type II cleavage only with Col2:3/4C(longmono) (C2C), type II synthesis (C-propeptide), the C-telopeptide of type II (Col2CTx), aggrecan 846 epitope, and cartilage oligomeric matrix protein (COMP). We performed a logistic regression to examine the relation of levels of each biomarker to the risk of cartilage loss in any knee. All analyses were adjusted for gender, age, and body mass index (BMI); results stratified by gender gave similar results. One hundred thirty-seven patients with symptomatic knee OA were assessed. At baseline, the mean (standard deviation) age was 67 (9) years and 54% were male. Seventy-six percent of the subjects had radiographic tibiofemoral OA (Kellgren & Lawrence grade of greater than or equal to 2) and the remainder had patellofemoral OA. With the exception of COMP, none of the other biomarkers was a statistically significant predictor of cartilage loss. For a 1-unit increase in COMP, the odds of cartilage loss increased 6.09 times (95% confidence interval [CI] 1.34 to 27.67). After the analysis of COMP was adjusted for age, gender, and BMI, the risk for cartilage loss was 6.35 (95% CI 1.36 to 29.65). Among subjects with symptomatic knee OA, a single measurement of increased COMP predicted subsequent cartilage loss on MRI. The other biochemical markers of cartilage synthesis and degradation do not facilitate prediction of cartilage loss. With the exception of COMP, if changes in cartilage turnover in patients with symptomatic knee OA are associated with cartilage loss, they do not appear to affect systemic biomarker levels.