Intersections, as the critical elements and the major bottleneck points of urban street networks, may have inconsistent performances in different countries. This is largely due to the fact that the factors affecting their performance e.g. driving behavior, vehicle characteristics, control methods, and environmental conditions may vary from one country to another. It is, therefore required to take into account these factors when developing or applying available models and methodologies for their capacity analysis or signal control setting. This is particularly important for the countries with heterogeneous and weak discipline traffic streams such as Iran. Meanwhile, estimating the saturation flow rate, which is a key parameter in capacity and delay analysis and in optimal timing of traffic signals, is of great importance. In this study, the possibility of identifying and or developing appropriate models for estimating the saturation flow rate at the signalized intersections in these situations has been explored. For this purpose, a case study performed at the signalized intersections located in the city of Yazd, a medium sized city located in the middle of Iran. Using the data obtained from several intersections together with the application of analytical procedures proposed by American, Australian, Canadian, Indonesian, Iranian and Malaysian highway capacity guides, the saturation flow rate was estimated from both field observations and analytical methods. A comparison of these results indicated that in the protected left-turn situations, the Australian guide produced the best comparable results with the field data. On the other hand, in the permitted left-turn situations, the method proposed in the American Highway Capacity Manual guide produced the best comparable results with the field data. Furthermore, three new models were developed for estimating the saturation flow rate in three different situations namely, unopposed mixed straight and turning traffic movements, opposed mixed straight and turning traffic movements and merely straight through movement. The effective width, traffic composition, and opposite oncoming through traffic flow were considered as the effective parameters in the proposed models. Moreover, using the multivariate regression analysis, the Passenger Car Equivalent coefficients for motorcycles and heavy vehicles were calculated as 0.51 and 2.09, respectively.