Serum proteomic analysis is an analytical technique utilizing high-throughput mass spectrometry (MS) in order to assay thousands of serum proteins simultaneously. The resultant 'proteomic signature' has been used to differentiate benign and malignant diseases, enable disease prognosis, and monitor response to therapy. This pilot study was designed to determine if serum protein patterns could be used to distinguish patients with tumour-stage mycosis fungoides (MF) from patients with a benign inflammatory skin condition (psoriasis) and/or subjects with healthy skin. Serum was analysed from 45 patients with tumour-stage MF, 56 patients with psoriasis, and 47 controls using two MS platforms of differing resolution. An artificial intelligence-based classification model was constructed to predict the presence of the disease state based on the serum proteomic signature. Based on data from an independent testing set (14-16 subjects in each group), MF was distinguished from psoriasis with 78.6% (or 78.6%) sensitivity and 86.7% (or 93.8%) specificity, while sera from patients with psoriasis were distinguished from those of nonaffected controls with 86.7% (or 93.8%) sensitivity and 75.0% (or 76.9%) specificity (depending on the MS platform used). MF was distinguished from unaffected controls with 61.5% (or 71.4%) sensitivity and 91.7% (or 92.9%) specificity. In addition, a secondary survival analysis using 11 MS peaks identified significant survival differences between two MF groups (all P-values <0.05). Serum proteomics should be further investigated for its potential to identify patients with neoplastic skin disease and its ability to determine disease prognosis.