Smoking behavior may be more persistent among those who expect that smoking will relieve negative affect (NA). Assessing smoking expectancies temporally close to mood situations could enhance the predictive value of that assessment. Dependent smokers (n = 71; 43 male, 28 female) participated in five laboratory sessions, each involving mood induction. The NA relief scale of the Smoking Consequences Questionnaire-Adult (SCQ-A), a very common measure of smoking expectancies during hypothetical situations, was assessed during initial screening. The SCQ-A was compared with a modified acute version administered each session, in which items asked about immediate expectancy for NA relief by smoking "right now" (termed Immediate Negative Affect Relief, or INAR). Actual NA relief due to smoking was measured each session by change on the NA scale of the Diener & Emmons Mood Form. The five sessions (counterbalanced) involved three different negative mood tasks, the negative mood condition of overnight smoking abstinence, and neutral mood (control). Generalized estimating equations showed that temporal proximity to the mood situation slightly enhanced the ability of expectancy to predict actual change in NA due to smoking, as the interaction with condition was significant for the INAR but marginal for the SCQ-A. However, the acute INAR predicted NA relief due to smoking only after overnight smoking abstinence and not during the other specific mood induction conditions, contrary to expectations, while the SCQ-A was not significant during any of the individual conditions. In sum, assessment of expectancy for NA relief may be of limited use in predicting actual NA relief from smoking during a current mood situation, aside from NA due to overnight abstinence.