Background/Purpose:In the phase 3 TENDER trial of tocilizumab (TCZ) in patients with systemic juvenile idiopathic arthritis (sJIA), decreases in neutrophil count were commonly observed. The purpose of this study was to determine whether neutropenia was associated with increased risk for infection and to investigate variables associated with the development of neutropenia in patients treated with TCZ for up to 2 years in TENDER.Methods:One hundred twelve children with active, persistent sJIA were randomly assigned 2:1 to receive TCZ based on body weight (12 mg/kg <30 kg or 8 mg/kg ≥30 kg) or placebo intravenously every 2 weeks for 12 weeks and continued in an ongoing, TCZ openlabel extension (). Worst Common Toxicity Criteria (CTC) neutropenia grade (grade 1, ≥1.5 and <2.0 × 109/L; grade 2, ≥;1.0 and <1.5 × 109/L; grade 3, ≥0.5 and <1.0 × 109/L; grade 4, <0.5 × 109/L) and lowest observed neutrophil count (109/L) were identified for each patient. Univariate linear regression analysis was performed to investigate the association of patient characteristics with lowest observed neutrophil count. Rates of infection and serious infection (per 100 patient‐years [PY]) in periods ±15 days around grade 1–2 neutropenia (22.9 PY) and around grade 3–4 neutropenia (5.5 PY) were compared with corresponding rates in periods with normal neutrophil count (173.6 PY).Results:Up to week 104, 64 of 112 patients (57.1%) had at least 1 episode of grade 1–4 neutropenia; worst grade: 1 (n = 2), 2 (n = 34), 3 (n = 26), and 4 (n = 2). Rates of infection and serious infection during periods of normal neutrophil counts (276.5/100 PY [95% CI: 252.3, 302.3] and 11.5/100 PY [95% CI: 7.0, 17.8], respectively) were comparable with those observed ±15 days around grade 1–2 neutropenia (226.7/100 PY [95% CI: 169.3, 297.3]; 8.7/100 PY [95% CI: 1.1, 31.5]) and grade 3–4 neutropenia (292.5/100 PY [95% CI: 167.2, 475.0]; 0/100 PY), with no trend toward increased risk with higher grade neutropenia. Methotrexate use (Yes/No) was significantly associated with lowest observed neutrophil count (difference: −0.575 [95% CI: −1.02, −0.13], p = 0.012), with 62% of 77 patients receiving methotrexate versus 46% of 35 patients not receiving methotrexate having grade 1–4 neutropenia. Younger age was borderline associated with lowest observed neutrophil count (β = 0.04661; p = 0.047). Concurrent use of glucocorticoids and TCZ exposure were not associated with lowest observed neutrophil count (p > 0.3).Conclusion:No trend for association between neutropenia and increased risk for infection was observed in the TENDER trial. Background methotrexate, and somewhat younger age, was associated with increased risk for neutropenia, whereas TCZ exposure and concurrent glucocorticoid use were not.