Horizontal and vertical elements of a highway are designed based on an assumed design speed. This concept was developed in the 1930s as a mechanism for designing rural alignments to permit most drivers to operate uniformly at their desired speed. In 1938, AASHO recognized that drivers select a speed influenced by the roadway environment instead of an assumed design speed. Recent research suggests that design speed is no longer the speed adopted by the faster group of drivers but that it has become a value used to establish the sharpness of horizontal and vertical design elements. The objective of this study was to establish the relationship between design and operating speeds for crest vertical curves with limited sight distance. Geometric data and 3,500 paired speeds (speeds at control and crest sections) were collected at 36 sites in 3 states. The results indicated that both the 85th percentile and the mean operating speeds were well above the inferred design speeds of the crest vertical curves for the range of conditions studied and that the lower the design speed the larger the difference between the 85th percentile speed and the design speed. The mean reductions in speed between the control and crest sections tend to increase as available sight distance is decreased; however, the reduction in speed is less than that suggested by current AASHTO criteria.