Neurodegenerative diseases involve disruption of healthy brain network communication occurring before the emergence of symptoms. Magnetoencephalography (MEG) is sensitive to the magnetic fields generated by cortical neuronal activity, and is the most spatio-temporally accurate method of directly assessing neuronal activity non-invasively. We used MEG to directly compare three neurodegenerative disorders with a large healthy cohort to characterise patterns of activity deviating from healthy ageing. Task-free MEG recordings were acquired from patients with Alzheimer's disease (AD, n = 29), Parkinson's disease (PD, n = 25), amyotrophic lateral sclerosis (ALS, n = 33) and healthy controls (HC, n = 191). Healthy ageing trajectories for metrics including spectral power (local neuronal recruitment), connectivity (long-range communication), 1/f exponent (power spectrum slope, which may reflect inhibition), and oscillatory speed were extracted. These metrics were compared pairwise between HC and patient groups, controlling for age and sex. The modelled trajectories of healthy ageing included increasing beta power and oscillatory speed, with reduced power spectrum slope. PD, AD, and ALS groups all showed reductions in beta power and slowing of oscillatory activity compared to matched HC. In AD, older patients showed lower beta power compared with younger patients. Compared with matched HC, the power spectrum slope was uniquely reduced in ALS, in contrast to the increase seen in PD and AD. Gamma connectivity increased in AD and ALS. MEG has unique potential as a source of biomarkers that might be used to detect deviation from healthy ageing if applied at an earlier presymptomatic stage of neurodegeneration than current tools permit. It might also provide outcome measures for prevention trials.