the major street and that on the minor street. That is a departure from MUTCD’s multitude of warrants and is, perhaps, an HCM perspec- tive that often emphasizes measures such as volume-to-capacity ratio and delay. The purpose of this paper, then, is to analytically verify this construct from an HCM perspective. That is, average control delay, the choice measure of effectiveness of HCM 2000 for inter- sections, will be analyzed and compared under various control types. Although there are other analytical and simulation tools capable of esti- mating delay at intersections, this study will use Highway Capacity Software (HCS) to stay true to the HCM perspective. The next section will provide some background information based on selected studies on delay at signalized as well as at stop-controlled intersections. HCM’s analytical models on control delay at intersec- tions with signal, AWSC, and TWSC will then be briefly reviewed. Readers already familiar with this subject may want to skip directly to the section on the design of case scenarios, which is followed by the analyses and results of these cases. A discussion on the results is presented at the end with some recommendations. BACKGROUND ON CONTROL DEVICE DELAY AT INTERSECTIONS Because this study focuses on delay, as opposed to a broader range of issues such as crash experience, signal coordination, and roadway network as identified by MUTCD’s signal warrants, a brief and far- from-inclusive introduction to the subject of delay calculations at variously controlled intersections is attempted here. For HCM 2000 and MUTCD, delay is a very important measure for evaluating the performance of an intersection. Even though the threshold values for different levels of service (LOS) are not identi- cal for different control types, control delay is the cornerstone of LOS for signal-controlled and stop-controlled intersections. Many models (1, 3, 4) have been developed over time for the estimation of control delay at signalized intersections. According to Han and Li (5), these models all project a lower (and thus desirable) delay level when a shorter cycle is used (that is, as long as the cycle length is not so short as to cause an undercapacity situation, leading to a dramatic increase in average delay). McKinley (6) suggests that one condition for installing a traffic control signal is that an all-way-stop controlled (AWSC) intersection must have experienced increased delay and congestion. He also observed that the safest intersection control, assuming reasonable compliance with the law, is the AWSC, even though it is also the most inefficient type of intersection control in most cases. Sampson (7 ) proposed the 4Q/6Q warrant to justify a signal. He argued that the warrant based on queue length is sensitive to a wide range of Control-Type Selection at Isolated Intersections Based on Control Delay Under Various Demand Levels Lee D. Han, Jan-Mou (James) Li, and Thomas Urbanik II 109 The Highway Capacity Manual (HCM 2000) displayed the figure Exhibit 10-15 for the purpose of forecasting the likely intersection con- trol types for future facilities. Because this figure is from a source exter- nal to HCM, to verify it, this paper employs HCM methodologies for the estimation and comparison of control delay, the choice measure of effec- tiveness at intersections controlled by signal, all-way stop signs, and two- way stop signs. After detailed analyses of more than 5,000 cases using Highway Capacity Software, results of control delay with various control types under a wide range of demand conditions were charted for compar- ison with Exhibit 10-15. It is found that Exhibit 10-15 is inconsistent with the results from HCM methodologies and, perhaps, should be replaced with the figures developed in this paper. On the basis of the criterion of minimizing delay alone, it is found that if demand is unbalanced between major and minor streets and if the traffic is low on minor streets, two- way-stop control should be used; if demand is somewhat balanced and minor streets see low to medium traffic, all-way-stop control is preferred; otherwise, signal control should be favored. The paper also demonstrates that the percentage of left-turning traffic has a significant effect on decisions involving intersection control types. With few exceptions, at-grade roadway intersections are controlled by traffic signals or stop signs for the allocation of right-of-way. For decades, the traffic signal warrants detailed in the Manual for Uniform Traffic Control Devices [MUTCD (1)] have served as the guiding principle for decisions about when to signalize an intersection. These warrants, which consider matters such as traffic demand and safety concerns, are to be used judiciously along with engineering studies and other local and practical considerations. In cases in which a traf- fic signal is not warranted or not installed, the intersection is usually controlled by two-way-stop control (TWSC) or all-way-stop control (AWSC). Chapter 10 of the 2000 edition of the Highway Capacity Manual (HCM 2000) provides some guidance so that “in the case of future facilities, the likely intersection control types can be forecasted using Exhibit 10-15” (2) (see Figure 1). This exhibit seems to suggest that the choice of proper intersection control types (i.e., signal, TWSC, or AWSC) can be determined on the basis of the two-way volume on L. D. Han, 112 Perkins Hall, and T. Urbanik II, 219B Perkins Hall, Department of Civil and Environmental Engineering, University of Tennessee, Knoxville, TN 37996- 2010. J.-M. (James) Li, University of Idaho, 115 Engineering Physics Building, Moscow, ID 83844-0901. Corresponding author: L. D. Han, lhan@utk.edu. Transportation Research Record: Journal of the Transportation Research Board, No. 2071, Transportation Research Board of the National Academies, Washington, D.C., 2008, pp. 109–116. DOI: 10.3141/2071-13