Rationale: Attentional bias toward drug-related stimuli is a feature of drug addiction that is linked to craving and drug-seeking behavior. Objectives/method: An attentional bias modification (ABM) program was tested in 42 methamphetamine-dependent clients (DSM-IV criteria) receiving residential treatment for their drug use. Participants were randomly assigned to one of two groups (N = 21 each), receiving 12 sessions of either computerized ABM training (designed to train attention away from methamphetamine stimuli 100% of the time) or an attentional control condition (designed to train attention away from methamphetamine stimuli 50% of the time). Outcome measures included attentional bias to methamphetamine-related stimuli on a probe detection task, self-reported craving, and preferences to view methamphetamine-related images on a Simulated Drug Choice Task. A subset of participants (N = 17) also underwent fMRI in a cue-induced craving paradigm. Results: Poor split-half reliability was observed for the probe detection task. Using this task, attentional bias toward methamphetamine-related stimuli was greater after training than at baseline, irrespective of group (p = 0.037). Spontaneous and cue-induced methamphetamine craving diminished with time (ps < 0.01), but ABM training did not influence these effects (group by time interactions, ps > 0.05). ABM training did not influence selection of methamphetamine-related pictures in the Simulated Drug Choice task (p > 0.05). In the fMRI assessment, cue-induced activation in the ventromedial prefrontal cortex was reduced over time, without an effect of ABM training. Conclusions: ABM training did not improve several clinically relevant variables in treatment-seeking methamphetamine users. Additional research is needed to improve the measurement of attentional bias.