A new triage test for human blood samples can distinguish active cases of tuberculosis (TB) from similar diseases in adults in less than an hour - helping to meet an elusive goal for global health authorities. The test is more sensitive than previous versions and may offer a point-of-care method for spotting TB cases, which often go undiagnosed in vulnerable patients such as those with HIV. The prevention and control of TB depends on having an effective, accessible test to detect cases and identify people who are at risk of infection. However, diagnosing cases of active TB can be difficult because there are many other diseases that cause TB-like symptoms. This challenge, on top of obstacles such as drug resistance and a high burden in the HIV-positive population, has thwarted efforts to bring the disease in check, especially in developing countries. Seeking a solution, Rushdy Ahmad and colleagues studied three cohorts of patients with chronic cough (406 total) using machine learning techniques and identified four blood proteins that could distinguish active cases of TB from TB-like diseases. The research team then created an ultra-sensitive immunoassay that screens for these proteins in blood samples. Their panel could pick out TB infections in 317 samples from patients with persistent cough from Africa, Asia and South America, and performed well in patients regardless of their HIV infection status. Their assay was even more accurate when they added a fifth marker to the panel that detected antibodies against a bacterial antigen, showing a sensitivity and specificity of 86% and 69%, respectively. Although more work is needed, the assay approaches the performance guidelines set by the World Health Organization, and could be further developed into a workable test to triage patients.