Unlike antibody-mediated rejection (AMR) with clinical features, it remains unclear whether subclinical AMR should be treated, as its effect on allograft loss is unknown. It is also uncertain if AMR's effect is homogeneous across donor (deceased/live) and (HLA/ABO) antibody types. We compared 219 patients with AMR (77 subclinical, 142 clinical) to controls matched on HLA/ABO-compatibility, donor type, prior transplant, panel reactive antibody (PRA), age and year. One and 5-year graft survival in subclinical AMR was 95.9% and 75.7%, compared to 96.8% and 88.4% in matched controls (p = 0.0097). Subclinical AMR was independently associated with a 2.15-fold increased risk of graft loss (95% CI: 1.19-3.91; p = 0.012) compared to matched controls, but not different from clinical AMR (p = 0.13). Fifty three point two percent of subclinical AMR patients were treated with plasmapheresis within 3 days of their AMR-defining biopsy. Treated subclinical AMR patients had no difference in graft loss compared to matched controls (HR 1.73; 95% CI: 0.73-4.05; p = 0.21), but untreated subclinical AMR patients did (HR 3.34; 95% CI: 1.37-8.11; p = 0.008). AMR's effect on graft loss was heterogeneous when stratified by compatible deceased donor (HR = 4.73; 95% CI: 1.57-14.26; p = 0.006), HLA-incompatible deceased donor (HR = 2.39; 95% CI: 1.10-5.19; p = 0.028), compatible live donor (no AMR patients experienced graft loss), ABO-incompatible live donor (HR = 6.13; 95% CI: 0.55-67.70; p = 0.14) and HLA-incompatible live donor (HR = 6.29; 95% CI: 3.81-10.39; p < 0.001) transplant. Subclinical AMR substantially increases graft loss, and treatment seems warranted.