Visceral leishmaniasis (VL) is a fatal parasitic disease. Unfortunately, diagnosis of VL in east Africa currently relies on aspiration of tissue from the spleen or bone marrow, which is painful and potentially dangerous. We sought to determine whether peripheral blood could be used instead of invasive tissue aspirates to diagnose VL, using three parasite concentration techniques. Three hundred and one consecutive people suspected of having VL were recruited. Compared with microscopy of tissue aspirates, the diagnostic accuracy of peripheral blood microscopy was as follows: whole blood thin smear sensitivity 1.5% (95% confidence interval [CI] 0.0-8.3) and specificity 100% (95% CI 76.8-100); buffy-coat smear sensitivity 19.5% (95% CI 14.3-25.6) and specificity 98.9% (95% CI 94.1-100); peripheral blood mononuclear cell (PBMC) smear sensitivity 33.7% (95% CI 27.3-40.5) and specificity 95.7% (95% CI 89.6-98.6). Sensitivity of PBMC smears was significantly higher in Human immunodeficiency virus (HIV)-coinfected patients (N = 48/301); two-sample test of proportions, P = 0.0097; sensitivity 55.9% (95% CI 37.9-72.8) and specificity 92.9% (95% CI 66.1-99.8), and correlated with the degree of parasite load in the tissue. Combining the results from smears of both PBMC and buffy coat yielded a sensitivity and specificity of 67.6% (95% CI 49.1-82.6) and 92.9% (95% CI 66.1-99.8), respectively, in HIV-coinfected patients. In this setting, VL could be ruled-in with peripheral blood microscopy in a substantial number of VL suspects and may reduce the number of tissue aspirations performed, particularly in HIV-coinfected patients. More sensitive and logistically feasible methods than light microscopy are needed to detect Leishmania donovani parasites present in blood.