Objective To determine the relationship between serum levels of S100A8/A9 and S100A12 and the maintenance of clinically inactive disease during anti–tumor necrosis factor (anti‐TNF) therapy and the occurrence of disease flare following withdrawal of anti‐TNF therapy in patients with polyarticular forms of juvenile idiopathic arthritis (JIA). Methods In this prospective, multicenter study, 137 patients with polyarticular‐course JIA whose disease was clinically inactive while receiving anti‐TNF therapy were enrolled. Patients were observed for an initial 6‐month phase during which anti‐TNF treatment was continued. For those patients who maintained clinically inactive disease over the 6 months, anti‐TNF was withdrawn and they were followed up for 8 months to assess for the occurrence of flare. Serum S100 levels were measured at baseline and at the time of anti‐TNF withdrawal. Spearman's rank correlation test, Mann‐Whitney U test, Kruskal‐Wallis test, receiver operating characteristic (ROC) curve, and Kaplan‐Meier survival analyses were used to assess the relationship between serum S100 levels and maintenance of clinically inactive disease and occurrence of disease flare after anti‐TNF withdrawal. Results Over the 6‐month initial phase with anti‐TNF therapy, the disease state reverted from clinically inactive to clinically active in 24 (18%) of the 130 evaluable patients with polyarticular‐course JIA; following anti‐TNF withdrawal, 39 (37%) of the 106 evaluable patients experienced a flare. Serum levels of S100A8/A9 and S100A12 were elevated in up to 45% of patients. Results of the ROC analysis revealed that serum S100 levels did not predict maintenance of clinically inactive disease during anti‐TNF therapy nor did they predict disease flare after treatment withdrawal. Elevated levels of S100A8/A9 were not predictive of the occurrence of a disease flare within 30 days, 60 days, 90 days, or 8 months following anti‐TNF withdrawal, and elevated S100A12 levels had a modest predictive ability for determining the risk of flare within 30, 60, and 90 days after treatment withdrawal. Serum S100A12 levels at the time of anti‐TNF withdrawal were inversely correlated with the time to disease flare (r = −0.36). Conclusion Serum S100 levels did not predict maintenance of clinically inactive disease or occurrence of disease flare in patients with polyarticular‐course JIA, and S100A12 levels were only moderately, and inversely, correlated with the time to disease flare.