A sample of 535 entrants to opioid dependence treatments across three treatment modalities were administered a structured interview to ascertain the prevalence of non-injecting heroin use. Ten per cent of participants had used heroin primarily by smoking/inhaling in the month preceding interview, and 9% had used heroin and other drugs exclusively by non-injecting routes. Non-injectors were younger (25.3 vs. 29.5 years), had higher levels of eduction (10.6 vs. 10.0 years), were more likely to be employed (33 vs. 18%) and had lower levels of recent crime (31 vs. 56%). They also had shorter heroin using careers (5.1 vs. 9.9 years), fewer symptoms of dependence (5.1 vs. 5.6), had been enrolled in fewer previous treatment episodes (3.3 vs. 11.5) and had less extensive lifetime (8.0 vs. 9.1 drug classes) and recent (3.6 vs. 4.9) polydrug use. Non-injectors were substantially less likely to report lifetime (13% vs. 58%) or recent (2% vs. 29%) heroin overdoses. There were no differences between the general physical and psychological health of the two groups. While non-injectors had a lower level of post-traumatic stress disorder (29% vs. 34%), there were no differences in levels of major depression, attempted suicide, antisocial personality disorder, or borderline personality disorder. A substantial minority of Australian treatment entrants are now using heroin exclusively by non-injecting routes. While this group is younger, and has substantially reduced risk of overdose and blood borne virus transmission, the physical and psychological health of non-injectors mirrors that of injectors.