INTRODUCTION: Mitochondrial DNA copy number (mtDNAcn), a measure of mitochondrial genomes per nucleated cell, has an unclear causal relationship with Alzheimer's disease (AD) and Parkinson's disease (PD). We integrated genetic correlation, polygenic risk scores (PRSs), and Mendelian randomization (MR) to assess whether mtDNAcn influences the risk of AD and PD, and evaluate how study-specific factors in mtDNAcn genome-wide association studies (GWASs) distort these causal estimates. METHODS: Using GWASs of four mtDNAcn measures, AD, AD/dementia, and PD, we evaluated genetic correlations, generated ancestry-normalized PRS in the Alzheimer's Disease Genetics Consortium (N = 27,383), and applied MR methods including latent heritable confounder-MR (LHC-MR). RESULTS: Across the four mtDNAcn GWASs, only one was consistently associated with AD/dementia and PD, with genetic correlations and PRSs showing negative correlations and MR indicating that higher mtDNAcn reduced AD/dementia and PD risk. DISCUSSION: Higher blood-based mtDNAcn was causally associated with reduced risk of AD/dementia and PD, with limited evidence to suggest a bidirectional effect.