The present study explores the applicability of the Many-Facet Rasch Measurement model to an Approach Avoidance Task assessing automatic approach tendencies toward alcohol. The MFRM was applied to 54 alcohol dependent outpatients who completed a combined Cognitive Bias Modification (CBM) training, targeting alcohol approach and attentional bias. Main objectives were to examine a) occurrence of change; effect of b) experimental conditions and c) gender; d) measurement status of the measure. Main results included a) no main effect of time, which only modulates effects of experimental condition on approach/avoid tendencies; b) double CBM, and to a lower degree double placebo, outper-formed the other conditions, while approach bias placebo/attentional CBM had a negative effect; c) no gender differences; d) the measure taps into general and drink-specific approach/avoid tendencies, is stable in time, and is slightly sensitive to stimuli context. Methodological and clinical implications of study results are further discussed.