The question of which cut-off point would be more appropriate in the selection of sub-clinical obsessive-compulsive (OC) samples is an important methodological issue that has not been formally addressed, hence prompting the current study. Three groups of sub-clinical OC subjects, scoring 1, 1.5, and 2 standard deviations (SD) above the mean on the Padua Inventory, and a matched non-OC control group, were compared on various clinical and personality measures. As expected, the three sub-clinical groups had higher scores than the non-OC group on measures of OC symptoms (including obsessional slowness), depression, anxiety, and on personality measures of neuroticism, psychoticism, and obsessional traits. However, no significant differences were observed between the three sub-clinical groups on most measures, although the 2 SD group had higher scores on neuroticism and obsessional personality traits. It is concluded that a wide range of selection criteria can yield similar results that are interpretable within the current theoretical models of Obsessive-Compulsive Disorder (OCD). It is suggested that subjects with a diagnosis of OCD should be carefully excluded from sub-clinical studies. [A Spanish abstract follows the references.]