OBJECTIVES: To determine if baseline characteristics, treatment efficacy, psychosocial outcomes or tolerability were associated with patient preference for sildenafil citrate (sildenafil) or tadalafil for treating erectile dysfunction (ED) in men naive to phosphodiesterase 5 inhibitor therapy. PATIENTS AND METHODS: In an open-label, crossover study of sildenafil (25, 50 or 100 mg) and tadalafil (10 or 20 mg), dosed as needed, after a 4-week baseline assessment, 367 men with ED were randomly assigned to sildenafil followed by tadalafil or vice versa (8-week dose optimization and 4-week assessment phase for each treatment period). Patients completing both periods chose which treatment they preferred for an 8-week extension phase. Bivariate logistic regression and stepwise logistic regression were used to determine if any baseline characteristics or post-baseline measurements were associated with the patients' treatment preference. Baseline variables examined were age, race, ED aetiology/duration, body mass index, smoking status, alcohol consumption, vital signs, comorbid medical conditions, and baseline scores for the International Index of Erectile Function (IIEF) domains, Psychological and Interpersonal Relationship Scales (PAIRS) domains, and Sexual Encounter Profile (SEP) diary questions. Post-baseline variables examined were therapy sequence, dosage, and differences in IIEF and PAIRS domains, SEP scores, in number/timing of sexual attempts and in the severity of side-effects (overall patient perception). RESULTS: Of 291 patients completing both treatments and indicating a preference, 85 (29%) preferred sildenafil and 206 (71%) preferred tadalafil. Variables were individually analysed using bivariate analysis; one baseline characteristic (presence/absence of hyperlipidaemia) and 13 post-baseline measurements were significantly associated with the patients' treatment preference. Variables were analysed as a group using stepwise logistic regression; a set of six post-baseline factors was identified as significantly associated with patient preference. Dosage choice, reductions in the PAIRS time concerns domain, IIEF intercourse satisfaction domain improvements, smaller side-effect severity scores, more sexual attempts, and increased SEP4 scores (satisfaction with erection hardness) during the tadalafil or sildenafil treatment periods were all significantly associated with preference for tadalafil or sildenafil. CONCLUSIONS: We identified no baseline characteristics that prospectively distinguish patients who will prefer tadalafil or sildenafil. Patient differences in time concerns, dosage choice, intercourse satisfaction, treatment tolerability, number of sexual attempts and satisfaction with erection hardness were the set of factors most significantly associated with treatment preference, and the preference observed for tadalafil (71%) or sildenafil (29%) might be substantially accounted for by differences in these factors during the tadalafil and sildenafil treatment periods.