To determine relatively useful diagnostic procedures of carbon disulfide (CS2) poisoning in terms of validity and cost-effectiveness, several diagnostic tests are evaluated on 1,552 people by prevalence ratio (exposed/nonexposed), dose-response relationship, sensitivity and specificity, and the cost of the tests. Several symptoms with high kappa indices are found to be useful in various combinations, showing a consistent dose-response relationship and high exposed-nonexposed ratio. In clinicopathologic tests on functions of the kidney, liver and hematopoietic systems, eight items out of 22 have been shown to have significant dose-response relationship, mostly in liver function tests. A thorough dental examination failed to identify any useful indicator peculiar to the CS2 poisoning. Unlike Western people, the prevalence of coronary heart disease among Koreans was too low to be useful in the diagnosis of CS2 poisoning. Among four elective tests, i.e., Minnesota Multiphasic Personality Inventory (MMPI), Brain magnetic resonance imaging (MRI), nerve conduction velocity (NCV), and fluorescent angiography (FAG), the NCV appeared to be a more sensitive and specific test than the others are. Combinations of the tests improved the probability of diagnosing CS2 poisoning cases when any one test out of four was positive. Addition of other valid tests increased the probability of excluding non-cases. It was concluded that diagnosis of CS2 poisoning could be made validly and inexpensively if the diagnostic tests were carefully chosen step by step.