Roundabout models, available in the literature widely, vary between one method and another. Majority of the\ud
models are solely based on circulating fl ow to estimate capacity. The relationship between maximum entry\ud
fl ow of large dual and triple lane roundabouts and their geometric and traffi c characteristics is investigated\ud
here for saturated fl ow condition using multivariate analysis. The developed model, based on the data gathered\ud
from 13 roundabouts in Bahrain, matched the fi eld data reasonably well. The signifi cant predictors, out of\ud
60 tested ones, included circulating and exiting fl ows, number of entry and circulating lanes, circulating and\ud
entry widths, inscribed diameter and fl are length. Capacities predicted through various international models\ud
varied considerably with both the developed model and the fi eld data. The international models were then\ud
calibrated against the model, and consequently the fi eld data, so as to have signifi cant match with the two. The\ud
procedure followed in developing the model and calibrating the international models consisted of eight stages.\ud
It involved identifi cation of signifi cant correlation of individual predictor with capacity, multivariate regression\ud
analysis, model smoothing, multi-collinearity test, redevelopment of the model after adjusting the variables\ud
causing collinearity, comparison analysis with seven known international models and calibration of the international\ud
model. While UK, aaSIDRA and French models required around 50% reduction to match the developed\ud
model and actual data; US FHWA and Swiss models required 25% reduction and US HCM and German models\ud
required just 8% adjustments. Such clear variations call for further research. The fi ndings assist the urban planners\ud
as when to shift from one type of intersection control into another involving roundabout, and vice versa