Abstract During slug flow, knowledge of the slug frequency is essential for the gas -liquid receiving facility design as well as for predicting various slug flow characteristics such as slug length and pressure drop. Various methods proposed in the literature for predicting the slugging frequency in horizontal and inclined pipes were examined. These included both empirical correlations as well as mechanistic models. Slug flow frequency data taken with an air-water system in a laboratory flow loop together with data from the published literature were compared to the predictions of the various methods. A total of 399 data points were collected covering pipe diameters from 1 to 8 inches and inclinations from 0 to 11 degrees above the horizontal. A total of eight published methods were compared to the data but none was found satisfactory. For this reason, the mechanistic slug frequency model of Taitel and Dukler (1976) was investigated in detail as an alternative unbiased prediction method. This model required the solution of the unsteady-state equations for mass and momentum by a finite difference technique. This numerical model gave satisfactory results at the expense of considerable computer CPU time. For faster slug frequency calculations a new correlation was developed utilizing all 399 data points. This resulted in 0% average error (bias) and 60% average absolute error. This correlation represents a significant improvement in slug frequency prediction accuracy over the other methods studied.