Despite high rates of comorbid substance use disorders and smoking among people with psychotic disorders, previous studies have not examined differences in socio-demographic, clinical or disability characteristics between psychosis sub-groups with different patterns of substance use. This study compared the characteristics of five groups of people with psychosis and varying patterns of substance use (n = 1152), including groups entering treatment studies for substance use or smoking, epidemiological samples of substance users and smokers, and people without such problems. Data were drawn from several recent Australian studies using comparable structured interviews and scales. There were moderate group differences in illness and symptom profiles, with substance users tending to have higher depression and reality distortion scores. Unexpectedly, personal disability was considerably lower among those seeking treatment for their substance use compared with the epidemiological groups, raising concerns about the appropriateness of traditional recruitment approaches in treatment studies and highlighting the need for more assertive treatment engagement and referral strategies in routine clinical settings. As a consequence of uncertainty about links between substance use and everyday functioning, it is suggested that health messages to young people may need to strike a better balance between the potential harms and benefits associated with substance use in pursuing broader harm minimization goals.