This study aimed to identify biomarkers to distinguish adult-onset Still's disease (AOSD) and to predict disease phenotypes. In total, 49 patients diagnosed with AOSD and 200 patients with common diseases (controls) were included in the analysis. The levels of 69 cytokines were analyzed using a multi-suspension cytokine array. Cytokine cluster analysis was performed to identify specific molecular networks. Furthermore, random forest analysis and logistic regression analysis were used to rank cytokines based on their importance and to determine specific biomarkers for identification of AOSD patients and phenotypes. Patients with AOSD demonstrated significantly higher macrophage migration inhibitory factor (MIF) and interleukin (IL)-12(p40) serum levels than controls and patients with rheumatoid arthritis. Serum levels of chemokine (C-C motif) ligand (CCL) 8 and CCL22 were significantly lower in AOSD patients with a polycyclic systemic disease phenotype and could be differentiated with high accuracy from the other phenotypes (cutoff value for CCL8 = 122.7 pg/mL, CCL22 = 593.3 pg/mL, sensitivity 66.7%, specificity 87.1%, area under the curve 0.843). Combined MIF and IL-12(p40) levels may represent a biomarker for differentiating patients with AOSD from those with other diseases. The chemokine profiles of AOSD with a polycyclic systemic disease phenotype may differ from other phenotypes.