Smoking withdrawal reports from a clinical trial (n = 893) were submitted to hierarchical linear modeling as a cross-method replication of a heterogeneity approach to withdrawal measurement and to clarify the influence of postcessation smoking on symptom reports. Five individual difference parameters tapping distinct facets of withdrawal were derived: intercepts (mean severity), linear slope (direction and rate of change), quadratic trend (curvature), volatility (scatter) and, among lapsers, a cigarette coefficent (smoking-related deflections of symptoms). All parameters were highly variable across persons. Lapsers had more aversive symptom patterns than abstainers, and symptoms tended to be higherthan otherwise predicted on lapse days. These results reinforce the conclusion that withdrawal symptoms are highly variable and argue against discarding withdrawal data from participants who lapse.