OBJECTIVES: To conduct a post-hoc analysis of the time-courses of MNWS item or domain scores (MNWS scores) and weekly point prevalence of abstinence (PVR) during the treatment phase of a previously-published (Aubin et al, 2008) randomized open-label clinical trial of varenicline (N=376) vs. NRT (N=370). METHODS: Current cigarette smokers, motivated to quit smoking, participated in the trial and completed the MNWS instrument.  Descriptive statistics (mean±standard error) of the MNWS scores from weeks 2 to 7 were computed.  Time-course comparisons stratified by PVR were performed, with weekly responders defined by PVR=0 and nonresponders otherwise. Multivariate repeated-measures mixed-effects regression was conducted for each MNWS domain score as the outcome variable.  Covariates included baseline, treatment, patient characteristics, smoking history and PVR.  Statistical significance was reached when two-sided p≤0.05.  RESULTS: The mean baseline MNWS scores of varenicline vs. NRT were comparable.  By PVR, varenicline significantly reduced the mean urge to smoke vs. NRT in weeks 2, 3, 4 and 5 (0.45±0.10; 0.28±0.09; 0.27±0.10; 0.27±0.09, respectively; all p<0.01) among responders, and in weeks 2 to 6 (0.59±0.13; 0.44±0.14; 0.46±0.16; 0.52±0.18; 0.38±0.17; all p<0.03) among non-responders.  The mean negative affect scores were significantly lower in weeks 2 to 5 and 7 (0.31±0.07; 0.20±0.06; 0.18±0.07; 0.23±0.07; 0.13±0.06; all p<0.03) among responders and in week 2 (0.22±0.11; p<0.04) among non-responders.  Additionally, restlessness was significantly reduced in weeks 2 to 5 (0.45±0.10; 0.33±0.09; 0.22±0.09; 0.22±0.08; all p<0.02) among responders and in week 2 (0.42±0.14; p<0.003) among non-responders, and similarly the reduction of increased appetite (0.38±0.18; p<0.05) among nonresponders in week 7.  CONCLUSIONS: Overall, lower mean patient-reported MNWS scores associated with symptoms of tobacco withdrawal were observed for varenicline than for NRT, reaching statistical significance, particularly among abstainers. Differences in the dynamics of treatment effects along with concomitant abstinence status warrant further bivariate analyses.