One hundred ninety-two patients with previously untreated metastatic cancer (102 non-small-cell lung cancer [NSCLC]; 90 colorectal cancer) were randomized to receive either ad lib nutritional intake (control group) or specific nutritional intervention during a 12-week study period when chemotherapy was administered. Those patients randomized to nutritional interventions were counselled to take oral nutrients with caloric intake equal to 1.7 to 1.95 times their basal energy expenditure, depending on their pretreatment nutritional status ("standard" group). An augmented group was counselled to have a caloric intake equivalent to that of the standard group but with 25% of calories provided as protein and additional supplements of zinc and magnesium. Counselling increased caloric intake in both tumor types but reduced weight loss in the short term only for lung cancer patients. Ninety-three NSCLC patients were evaluable for tumor response to vindesine and cisplatin. Overall, only 20.4% of the patients responded, and there were no significant differences in response rates, median time to progression, or overall duration of survival between the nutrition intervention groups and the control group. The tumor response rate to time-sequenced 5-fluorouracil (5-FU) and methotrexate in the 81 evaluable patients with colorectal cancer was only 14.8%, and no significant differences in tumor response rates were noted between the three groups. Furthermore, the median time to progression and overall duration of survival were not different for the control, standard, and augmented groups. Nutritional interventions using dietary counselling had no impact on the percent of planned chemotherapy dose administered, the degree of toxicity experienced by patients, or the frequency of treatment delays. A multivariate prognostic factor analysis demonstrated that for lung cancer, the percent of weight loss, serum albumin concentration, and presence of liver metastases were significant (P less than .05) and independent prognostic variables for survival duration. For colorectal cancer, serum albumin, alkaline phosphatase, lactic dehydrogenase (LDH) levels, and percent targeted caloric intake (TCI) were significant independent predictors of survival duration.