Experience with human responses to antitumor drugs and with cultures of 5-fluoro-2′-deoxyuridine sensitive N1S1 tumor cells exposed to the drug has demonstrated that several factors which are not evaluated in predictive tests may affect the clinical response of tumors to drugs. Such factors lead to discrepancies between the in vitro assessment of response and that actually obtained in the patient. Studies with the fluorinated pyrimidines indicate that the cells must be biochemically sensitive before response can occur. However, biochemical sensitivity may exist and yet no tumor regression occur when the drug is given to the patient. Tumor cells may be killed but their destruction may be masked in at least two ways: Killed, sensitive cells may be replaced by fibrous tissue. Or, drug-resistant cells present in the original tumor may rapidly assume dominance when the sensitive cells are killed. In either case, significant tumor response to drug may escape detection.