Cancer patients, assessed as nutritionally at risk, were randomly assigned to a control group or to one of four intervention groups receiving (a) nutritional supplementation, (b) relaxation training, (c) both nutritional supplementation and relaxation training, or (d) neither nutritional supplementation nor relaxation training. Fifty-five subjects completed a four-month intervention period during which they were visited biweekly by a nurse (except control subjects). In repeated measures analyses of variance, significant group-by-time interactions were obtained for weight and arm muscle circumference (a measure of protein stores), indicating that for these measures the groups changed differentially during the intervention period. The group-by-time interaction approached significance on the Karnofsky Performance Status Scale. For all three variables, gain was greatest for the relaxation group; the most severe loss occurred in the control group. These findings suggest that the cachexia of cancer may be slowed or reversed through noninvasive nursing interventions.