Despite of evidence suggesting that the route of administration should be taken into account to assess the short and long-term effects of cocaine consumption, to our knowledge no study to date has characterized clinically relevant neuropsychological and physiological variables comparing individuals with histories of smoked cocaine dependence (SCD) and insufflated cocaine hydrochloride dependence (ICD). The present study examined a sample of 25 participants who fulfilled criteria for SCD, 22 for ICD, and 25 healthy controls matched by age, gender, education, and socioeconomic status. An exhaustive NPS battery was used to assess cognitive domains (attention, executive functions, fluid intelligence, memory, language and social cognition). We complemented this NPS assessment with structural (MRI) and functional (fMRI) neuroimaging data. We found that different routes of administration led to equally different profiles of neurocognitive impairment, with the SCD group being specifically associated with deficits in executive-attention functions. Consistent with risk models, executive-attention function deficit is better explained by age and age at the first use of the drug. SCD presented reduced grey matter density relative to ICD and CTR in the bilateral caudate, a key area for executive functions and attention. Specifically, connectivity between left caudate and inferior frontal regions in SCD mediated performance-structure association.