Abstract—One tool employed by various state transportation departments for traffic queue prediction, given the specifics of a proposed freeway work zone, is the Excel-based “Lane Rental Model” developed at the Oklahoma Department of Transportation (OkDOT), using the 1994 Highway Capacity Manual (HCM) lane capacity tables. Preliminary testing of the OkDOT tool confirmed lack of accuracy. Logic errors were corrected to form a baseline, and two other versions were created using the lane capacity model of HCM 2000: an HCM 2000 version using work intensity effects of -160 to +160 passenger cars per hour per lane (pcphpl); and an HCM 2000 hybrid version using work intensity penalties of -500 to 0 pcphpl. Using a diverse set of 32 actual freeway work zone lane closure descriptions as the “test data bank,” we compared predictions produced by the three versions of the OkDOT spreadsheet tool with the actual Maximum Queue Length (MQL) observed in the field. The HCM 2000 hybrid version with passenger car equivalent PCE = 2.1 for heavy vehicles is highly accurate, and minimized the overall error in predicting MQL. The “empirical optimization” methodology used to reach this conclusion is the major contribution.