The Judgment of Line Orientation (JLO) test was developed to be, in Arthur Benton's words, "as pure a measure of one aspect of spatial thinking, as could be conceived" (Benton, 1994, p. 53). The JLO test has been widely used in neuropsychological practice for decades. The test has a high test-retest reliability (Franzen, 2000), as well as good neuropsychological construct validity as shown through neuroanatomical localization studies (Tranel, Vianna, Manzel, Damasio, & Grabowski, 2009). Despite its popularity and strong psychometric properties, the full-length version of the test (30 items) has been criticized as being unnecessarily long (Straus, Sherman, & Spreen, 2006). There have been many attempts at developing short forms; however, these forms have been limited in their ability to estimate scores accurately. Taking advantage of a large sample of JLO performances from 524 neurological patients with focal brain lesions, we used techniques from item response theory (IRT) to estimate each item's difficulty and power to discriminate among various levels of ability. A random item IRT model was used to estimate the influence of item stimulus properties as predictors of item difficulty. These results were used to optimize the selection of items for a shorter method of administration that maintained comparability with the full form using significantly fewer items. This effectiveness of this method was replicated in a second sample of 82 healthy elderly participants. The findings should help broaden the clinical utility of the JLO and enhance its diagnostic applications.