The Greek version of the Oldenburg Burnout Inventory, used to assess exhaustion and disengagement from work, was analyzed with several polytomous IRT models, namely the partial credit model, the generalized partial credit model, the rating scale model and the graded response model, to investigate the psychometric properties of the measure’s items. The OLBI was administered to 617 Greek employees, from which 314 participated also to the validation study. Results indicated that four negatively-keyed items from the original measure exhibited excellent psychometric properties (item/test information functions) and were used for the screening version of the OLBI. The scr-OLBI was tested for differential item functioning (DIF) between male and female employees; no bias detected in relation to gender. Our results reveal that the scr-OLBI is a reliable and valid indicator of work-related burnout, which appears to be functionally equivalent to the original version, yet pertains the advantages of a screening tool.