Among the large variety of attentional tasks that have been used to study sustained attention, the Continuous Performance Task (CPT) is perhaps the most widely used. Despite substantial differences in task characteristics and demands, all CPT paradigms have been referred to as measures of sustained attention. In the present study we introduce a new variant of CPT, which minimizes perceptual and memory components while maximizing the sustained attention components of the task. In addition, we tested the contention that the ability to sustain attention should not be overly dependent on the specific stimuli and task-modality. To this end, we used a new visual Conjunctive CPT (CCPT) developed by Tsal, Shalev, & Mevorach (2005) and its auditory analogue. Using a Multi-Trait-Multi-Method (MTMM) analysis investigating reliability coefficients, convergent validity coefficients and divergent (discriminant) validity coefficients, we established that the new CCPT is a valid measure of sustained attention. In particular, high correlations were obtained between mean RT and SD-RT within each sensory modality. High correlations were also found between performance across sensory modalities (mean convergent validity: 0.71). Finally, low correlations (mean correlation of 0.18) were found between performance in the CCPT tasks and performance in two additional visuospatial attention tasks, which do not rely on sustained attention. These findings suggest that independent of sensory modality, the two CCPT tasks used here tap the same stable construct, namely, sustained attention. We conclude that when a measure of sustained attention is required, researchers should be aware of the caveats of standard CPTs and should be careful in selecting a proper task. Moreover, clinicians should appreciate that some CPT measures may reflect a combination of different cognitive operations rather than pure sustained attention. (Contains 3 figures and 2 tables.)