Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease characterised by functional connectivity alterations in both motor and extra-motor brain regions. Within the framework of network analysis, fingerprinting represents a reliable approach to assess subject-specific connectivity features within a given population (healthy or diseased). Here, we applied the Clinical Connectome Fingerprint (CCF) analysis to source-reconstructed magnetoencephalography (MEG) signals in a cohort of seventy-eight subjects: thirty-nine ALS patients and thirty-nine healthy controls. We set out to develop an identifiability matrix to assess the extent to which each patient was recognisable based on his/her connectome, as compared to healthy controls. The analysis was performed in the five canonical frequency bands. Then, we built a multilinear regression model to test the ability of the "clinical fingerprint" to predict the clinical evolution of the disease, as assessed by the Amyotrophic Lateral Sclerosis Functional Rating Scale-Revised (ALSFRS-r), the King's disease staging system, and the Milano-Torino Staging (MiToS) disease staging system. We found a drop in the identifiability of patients in the alpha band compared to the healthy controls. Furthermore, the "clinical fingerprint" was predictive of the ALSFRS-r (p = 0.0397; beta = 32.8), the King's (p = 0.0001; beta = -7.40), and the MiToS (p = 0.0025; beta = -4.9) scores. Accordingly, it negatively correlated with the King's (Spearman's rho = -0.6041, p = 0.0003) and MiToS scales (Spearman's rho = -0.4953, p = 0.0040). Our results demonstrated the ability of the CCF approach to predict the individual motor impairment in patients affected by ALS. Given the subject-specificity of our approach, we hope to further exploit it to improve disease management.