The goal of this research was to determine the effects of geometric characteristics on double left-turn lane (DLTL)operations, as measured with saturation (low rate (SFR) in units of passenger cars per hour of green per lane (pcphgpl). Data from 26 sites located in three states (Arizona, California, and Texas) were used in the analyses. The average 1)1.11. SFR was 1,775 pcphgpl. The inside-outside lane variable was found to be not significant, which means that the SFRs of the inside and outside lanes are similar. For each additional U-turning vehicle in the left-turn queue, SFR would decrease by 56 pcphgpl after accounting for the effects of other variables in the model. The model results indicate that the addition of a new lane on the receiving leg from a channelized right turn resulted in an SFR increase of about 50 pcphgpl. Width of the left-turn lanes did not significantly affect the SFR. The SFR determination method requires elimination of a queue if a heavy vehicle is present. Therefore, even though the operations of queues with only passenger cars are similar for the various left-turn lane widths studied (9.5 to 13 ft), results for the operations of queues that include heavy vehicles (i.e., trucks or buses) may be different. The width of the receiving leg represents the visual target for left-turning drivers. When the receiving leg width was between 24 and 36 ft, the average SFR was 1,725 pcphgpl, whereas a receiving leg width of 40 to 54 ft was associated with an average SFR of 1,833 pcphgpl. One potential change to practice on the basis of this research is to modify guidance documents to reflect the finding that the capacity of DLTLs is approximately 196% of a single median lane rather than 180%.