Working alliance and empathy are believed to be important components of counseling, although few studies have empirically tested this. We recently conducted a randomized controlled trial in which brief motivational and reduction counseling failed to increase the number of participants who made a quit attempt (QA) in comparison to usual care (i.e., brief advice to quit). Our negative findings could have been due to nonspecific factors. This secondary analysis used a subset of participants (n = 347) to test (a) whether, in comparison to usual care, brief telephone-based motivational or reduction counseling predicted greater working alliance or empathy; (b) whether changes in these nonspecific factors predicted an increased probability of a QA at a 6-month follow-up; and (c) whether counseling affected the probability of a QA via working alliance or empathy (i.e., mediation). Findings were similar for both active counseling conditions (motivational and reduction) versus usual care. In comparison to usual care, active counseling predicted greater working alliance (p < .001) and empathy (p < .05). Greater working alliance predicted a greater probability of a QA (p < .001) but, surprisingly, greater empathy predicted a decreased probability of a QA (p < .05) at the 6-month follow-up. Working alliance (p < .001) and empathy (p < .05) mediated the active counseling’s effects on the probability of a QA. One explanation for our motivational and reduction interventions’ failure to influence QAs in comparison to usual care is that working alliance and empathy had opposing effects on quitting. Our analyses illustrate how testing nonspecific factors as mediators can help explain why a treatment failed.