The paper investigates the validity of promising surrogate measures by investigating the link between such measures and safety. Such measures, if shown to be valid, can be used in two ways: a) to evaluate a newly implemented measure for which data can be quickly assembled to estimate the effect on the safety surrogate and, by implication, the effect on expected crashes, and b) to estimate the safety effects of a proposed change by first estimating the effect on the safety surrogate and then relating that outcome to the expected change in crashes. The idea is not new. In fact, several researchers have been working on various complementary pieces of the puzzle. The key missing link is in developing measures that can in fact be linked explicitly to safety as manifested in expected numbers of crashes. In addressing this void, two complementary sets of models are explored – models to relate surrogate measures to design features and models to relate crashes to surrogate measures. The first set successfully modeled simulated conflicts for urban signalized intersections estimated from micro-simulation software, complementing the models for measured and predicted speeds for roundabouts that have been developed in our recently published research. Models relating these surrogate measures to observed crashes suggest that there is promise in the approach of using certain surrogate measures for crash-based safety estimation.