The latent inhibition (LI) paradigm has been used to assess attentional dysfunction in various pathological groups. The rationale is based on the assumption that passive stimulus exposure results in the acquisition of an inattentional response to that stimulus. Consequently, compared to a novel stimulus in the same new learning situation, the preexposed stimulus is at a disadvantage. It is argued that methodological and conceptual problems in constructing procedures and designs have created obstacles in relating disrupted LI to psychopathology. Specifically, issues associated with within- and between-subject designs, dichotomous dependent variables, ceiling effects, converging operations, and possible mis-attribution of the LI effect are addressed. Designs and data from several new human-LI paradigms, with normal, de novo Parkinson, and schizophrenic subjects are examined. Results from a multi-condition, within-subject visual search procedure suggest that LI, heretofore attributed only to a deficit in the stimulus preexposed group, may, in part, be due to enhanced performance in the nonpreexposed group. Implications for the design and interpretation of LI experiments, particularly with pathologic groups are discussed.