A method for determining the endogenous urinary excretion levels of both 3-hydroxyproline and 4-hydroxyproline that may be useful for cancer screening of the general population and at the workplace is evaluated in this report. The excretion levels of 3-hydroxyproline and 4-hydroxproline were estimated in 97 patients with cancer and in 99 patients with various nonmalignant diseases and were compared with those of 211 healthy persons. Measurable 3-hydroxyproline peaks (by amino acid autoanalyzer) were absent from 93 samples from 211 healthy persons (44%), 50 of 99 patients with nonmalignant disease (50%), and 10 of 96 patients with cancer (10%). The levels of both 3-hydroxyproline and 4-hydroxyproline in cancer patients were significantly higher than those in healthy persons (p < 0.001 and p < 0.01, respectively) and those in patients with nonmalignant diseases (p < 0.05 and p < 0.01, respectively). Cancer patients were classified into three groups according to grade of cancer growth and invasion. The sensitivity of 3-hydroxyproline was 44% and higher than that of 4-hydroxyproline for the detection of stage II cancers (no distant metastasis); the sensitivities of both hydroxyprolines for the detection of stage I (very early cancer) were low. The specificity of these assays for healthy persons and patients with nonmalignant disease was 96% and 92% for 3-hydroxyproline, and 97% and 79% for 4-hydroxyproline, respectively. Urinary 3-hydroxyproline level should be further investigated as a cancer screening method for healthy persons in the community or the workplace, but appears unlikely to detect many cancers in the earliest stages.