Purpose: Patient involvement in medical care has been linked to more positive clinical experiences for patients, including better illness understanding and perceived illness control, as well as greater expectations regarding condition improvement and satisfaction with care. Little is known, however, about why patients do or do not initiate information exchange with their providers. This study examines the effects of heart failure (HF) patients' informational preferences, medical decision-making preferences, perception of providers' facilitation of patient involvement, HF severity, and patients' sociodemographic characteristics on patients' perceived level of patient-initiated information exchange (PIE). Methods: We conducted a telephone survey of HF patients being treated by a Veterans Affairs primary care or cardiology provider. PIE and perceived provider facilitation of patient involvement during patients' last clinic visit were measured using two independently scored subscales of the Perceived Involvement in Care Scale (i.e., Patient Information and Doctor-Facilitation). Patients' health information preferences were assessed using the Health Opinion Survey. Decision-making preference was assessed using the Control Preferences Scale (high score=preference for greater provider control/lesser patient control). Sociodemographic characteristics were assessed via self-report and patients' B-type natriuretic peptide (BNP) levels were extracted from patients' medical records. Descriptive, correlational, and linear regression analyses were conducted. Results: The 90 participants were mostly elderly (mean=70.1 years; range=42-88), male (94.4%), white (85.6%), had New York Heart Association (NYHA) Class II disease (55.6%). The BNP levels varied widely (mean=580.6 pg/mL; range=104–5000 pg/mL). Variables associated with PIE included: informational preferences (r=.557; p<.001); perceived doctor-facilitation of patient involvement (r=.432; p<.001); medical decision-making preference (r =-.356; p=.001); and educational attainment (r=.227; p<.05); with disease severity, falling just outside the critical region, (r=-.197; p=.062). Linear regression analysis indicated that these five variables explained about 46% of the variation in PIE (R²=.459). Three variables demonstrated significant independent effects: patients' informational preferences (β*=.441; p<.001), perceived doctor-facilitation of patient involvement (β*=.343; p<.001), and HF severity (β*=-.182; p<.05). Conclusions: In HF care, a patient's preference for more information may be a stronger predictor of PIE than the provider's facilitation of patient involvement. That greater HF severity had a negative effect (independent of other variables, including preferences) on PIE during last clinic visit could indicate that PIE is less likely with patients who may be feeling poorly or who are later in the treatment course.