INTRODUCTION: Homeless populations have high rates of smoking and unique barriers to quitting. General cessation strategies have been unsuccessful in this population. Smoking reduction may be a good intermediate goal. We conducted a secondary analysis to identify predictors of smoking reduction in a cohort of homeless smokers enrolled in a 26-week randomized clinical trial (RCT) targeting smoking cessation. METHODS: Data are from an RCT comparing motivational interviewing counseling plus nicotine replacement therapy (NRT) to brief advice to quit (standard care) plus NRT among homeless smokers. Using bivariate analyses and multinomial logistic regression, we compared demographics, health and psychosocial variables, tobacco use, substance use, and NRT adherence among those who reported: quitting; reducing smoking by 50-99%; and not reducing smoking by 50%. RESULTS: Of 324 participants who completed 26-week follow-up, 18.8% and 63.9% self-reported quitting and reducing, respectively. Compared to those who did not reduce smoking, participants reporting reducing indicated higher baseline cigarette use (OR=1.08; CI:1.04-1.12) and menthol use (OR=2.24; CI:1.05-4.77). Compared to participants who reduced, participants reporting quitting were more likely to be male (OR=1.998; CI:1.00-3.98), experience more housing instability (OR=1.97; CI:1.08-3.59), indicate higher importance of quitting (OR=1.27; CI:1.041.55), have higher NRT adherence (OR=1.75; CI:1.00-3.06), and lower odds of reported illicit drug use (OR=0.48; CI:0.24-0.95). CONCLUSIONS: Over half of participants reduced smoking by at least 50%, indicating reduction is feasible among homeless smokers. Further research is required to understand the impact of reduction on future cessation attempts in homeless smokers. This study shows that reduction is achievable and may be a valid intermediate goal.