Variables derived from the health locus of control construct and the Health Belief Model were used to predict compliance with medical regimen in chronically ill patients (adult-onset diabetics, hypertensives, and pulmonary disease patients). Stepwise multiple regression analyses were conducted with predictor variables: age, socioeconomic status, Internal Health Locus of Control (HLC), powerful others HLC, chance HLC, perceived severity of illness, outlook on illness, experienced symptoms, satisfaction with treatment, family support, and support of others. Dependent variables included three measures of compliance with medication regimen and two measures of compliance with self-management regimen. Patient satisfaction was a significant predictor of medication compliance, and lack of symptoms was most strongly associated with self-management compliance. Limitations of cognitive variables in predicting compliance in many chronic disease patients are discussed.