Critical illness polyneuropathy (CIP) has been linked to neurocardiac dysfunction mediated by autonomic nervous system dysregulation, which increases mortality. We aimed to assess if heart rate variability (HRV) biofeedback could improve neurocardiac function in CIP. We randomly allocated (1:1) patients with electrophysiologically confirmed CIP undergoing early inpatient neurological rehabilitation to additional HRV or sham biofeedback over 14 days. We evaluated neurocardiac function via standard deviation of normal-to-normal intervals (SDNN) as the primary outcome, as well as HRV frequency domains, sympathetic cutaneous sudomotor and vasomotor functions and disability at baseline, post intervention and 4 weeks later. The study is registered on the German Clinical Trials Register (DRKS00028911). We included 30 patients with CIP (40% females, median [interquartile range] age 64.6 [56, 72] years). We observed an increase in SDNN and the predominantly parasympathetic high frequency domain post intervention (ß = 16.4, 95% confidence interval [CI] 0.2, 32.6 [p = 0.047] and ß = 1179.2, 95% CI 119.9, 2158.5 [p = 0.018]), which was sustained at the 4-week follow-up (ß = 25.7, 95% CI 6.0, 45.4 [p = 0.011] and ß = 25.7, 95% CI 6.0, 45.4 [p = 0.011]). Patients who underwent HRV biofeedback displayed a higher adjusted Barthel index, indicating less severe disability 4 weeks after the intervention compared to those in the sham group (ß = 23.3, 95% CI 5.5, 41.1 [p = 0.014]). Low frequency and sympathetic skin functions did not differ between groups (p = nonsignificant). Our study provides pilot data suggesting that, in patients with CIP, HRV biofeedback can improve neurocardiac function with a predominant effect on the parasympathetic nervous system and has a beneficial effect on functional recovery.