Many studies confirmed the efficacy and safety of continuous infusion of intrajejunal levodopa/carbidopa gel (CIILG) for advanced Parkinson's disease (PD). Although this treatment is widely used, definite inclusion/exclusion criteria do not exist. In this prospective open-label study, we evaluated the long-term outcome in 28 consecutive patients and sought to detect any predictive factor to identify the best candidates for CIILG therapy. The assessment was carried out routinely at baseline, after 6 months and every year with UPDRS III-IV, FOG Questionnaire, non-motor symptoms scale, PD questionnaire (PDQ-8), cognitive and psychiatric status evaluation (MMSE, FAB, NPI) and caregiver's quality of life. 17/28 patients reached the 24-month follow-up. A statistically significant beneficial effect was shown on motor complications in short- and long-term follow-up, also on axial symptoms like gait disturbances. A concomitant improvement in PDQ8 score was observed, with a parallel mild amelioration, but not significant, on Caregivers QoL. When classified according to their outcome on QoL, the only predictive positive factor was less severe at Neuropsychiatric Inventory (NPI) score at baseline. Considering the improvement in motor scores (duration of "off" period), the more advanced age was associated with a poorer outcome. Our results confirmed a sustained efficacy and safety in long-term follow-up and suggest that younger age at operation and absence or mild presence of psychiatric/behavioural symptoms could be considered valid predicting factors in selecting the best candidates for this efficacious therapy.