Sensitive immunoassays are required for troponin, a low-abundance cardiac biomarker in blood. In contrast to conventional (analog) assays that measure the integrated signal of thousands of molecules, digital assays are based on counting individual biomarker molecules. Photon-upconversion nanoparticles (UCNP) are an excellent nanomaterial for labeling and detecting single biomarker molecules because their unique anti-Stokes emission avoids optical interference, and single nanoparticles can be reliably distinguished from the background signal. Here, the effect of the surface architecture and size of UCNP labels on the performance of upconversion-linked immunosorbent assays (ULISA) is critically assessed. The size, brightness, and surface architecture of UCNP labels are more important for measuring low troponin concentrations in human plasma than changing from an analog to a digital detection mode. Both detection modes result approximately in the same assay sensitivity, reaching a limit of detection (LOD) of 10 pg mL(-1) in plasma, which is in the range of troponin concentrations found in the blood of healthy individuals.