Abstract: Objective To characterize mycosis fungoides （MF）-associated serum proteins by surface enhanced laser desorption/ionization time of flight-mass spectrometry （SELDI-TOF-MS） technique. Methods Serum samples were collected following informed consent from 15 patients with MF and 15 patients with chronic eczema or neurodermatitis. Serum protein profiles were detected by CM10 protein chip combined with SELDI-TOF-MS. The proteins differentially expressed between MF and chronic eczema or neurodermatitis were assessed by the Zhejiang University Cancer Institute - ProteinChip Data Analysis System （ZUCI- PDAS）. Results A total of 329 protein peaks were detected from these patients, significant difference was observed in only 30 protein peaks. The optimal diagnostic model was developed by support vector machine algorithm with two protein peaks at Mass/Charge （m/z） 3939 and 5909 respectively. The expression of protein peak at m/z 3939 was higher in chronic eczema and neurodermatitis than in MF, while that at m/z 5909 appeared to express in an opposite way. In simulation training, this model proved to be able to distinguish MF from chronic eczema and neurodermatitis with both the specifity and sensitivity being 100 percent. The leave-one-out cross-validation also revealed a specificity and sensitivity of 100 percent for this model in the comparison of MF with chronic eczema and neurodermatitis. Conclusion These results suggest that SELDI- TOF-MS technique combined with bioinformatics is highly specific and sensitive in the diagnosis of MF.