Abstract Background There is a need for sensitive, robust, and scalable analytical methods for accurate, early detection of disease using less invasive sampling methods. Often, smaller volumes and low marker concentrations present challenges to achieving sufficient sensitivity and accuracy with traditional immunoassay methods. This is particularly the case for measuring markers for Alzheimer's and other neurological diseases in blood, where levels are typically at much lower concentrations than in CSF. Here we introduce a novel ultrasensitive technology for flow-based single molecule detection in an optofluidic chip. We have developed an Ultrasensitive Detection Module (UDM) device and optofluidic cartridge that enable easy integration of this technology into existing immunoassay systems. UDM sensitivity data is shown for three prototype biomarker assays in comparison to the Lumipulse G1200 (Fujirebio Inc., Japan). We also present analytical validation data for three Alzheimer’s Disease plasma biomarker assays developed on a fully automated benchtop analyzer with integrated UDM detector. Methods Ultrasensitive detection is achieved by integrated low-loss waveguide in an optofluidic chip that enables maximal excitation of fluorescent molecules traveling through a narrow fluidic channel. The resulting high-intensity emission is detected as a single digital event, enabling individual molecules to be counted without need for amplification. Fluorescence-based immunoassays were developed for detection with the UDM using high performance monoclonal antibodies and reagents (Fujirebio Inc., Japan). Capture antibodies were immobilized on magnetic particles and detection antibodies conjugated to a fluorescent reporter construct (“reporter-Ab”). Multiple incubation and wash steps were followed by dissociation of immune complexes and separation of reporter-Ab from magnetic particles. Released reporter-Ab molecules were injected into the UDM device for digital detection. Signal to noise performance was evaluated in the UDM using reporter construct alone (no assay), as well as for prototype PSA, IL-6, and TSH assays, to establish baseline performance for analytical sensitivity. Plasma assays for pTau181, Aβ1-40, and Aβ1-42 were developed and optimized for the automated benchtop analyzer and validated for analytical sensitivity (LOD/LOQ), linearity, parallelism, precision and repeatability using clinical samples. Results Detection signal-to-noise testing of the reporter construct alone (no immunoassay) demonstrated up to 1000-fold higher sensitivity than the Lumipulse G1200. Prototype assays for PSA, TSH, and IL-6 showed up to 118-fold, 36-fold, and 80-fold higher signal to noise ratio, respectively, when compared to the Lumipulse G1200. Plasma assays for pTau181, Aβ1-40, and Aβ1-42 showed preliminary LLOQs calculated at 0.03 pg/mL, 0.17 pg/mL and 0.37 pg/mL, respectively. Conclusions We have developed a new single molecule counting platform able to detect neurological biomarkers at sub-pg/ml levels in plasma, with the potential to provide sensitive and accurate diagnostic testing of low abundance biomarkers in blood. Such technologies can create new clinical value through higher detection sensitivity and accuracy in areas like Alzheimer’s and other neurological diseases, supporting clinical research and translation to clinical practice.