The paper compares the results of extensive measurements of the capacity of nonpriority traffic streams at major/minor junctions and roundabouts with the predictions of models based on gap-acceptance theory. There are significant differences in the form of the relation between the capacity and the traffic flows in the priority streams. The simple gap-acceptance models are poor predictors of the capacity in the United Kingdom, for they seriously overpredict at low values of priority flow and underpredict at high values. At the lower values the gap-acceptance predictions can only be made to agree with the results of the empirical measurements by introducing a dependence of the “move-up” times on the value of the flows in the priority streams. The discrepancies at the higher values of priority flow are probably due to gap-forcing and priority reversal, both of which are commonly observed in the United Kingdom, but whose occurrence may vary from one country to another as a result of differences in driver behavior, legal sanctions, and enforcement policies. Several gap-acceptance models of roundabout capacities are compared and are found to be mutually inconsistent.