This document presents techniques tor constructing and evaluating railroad grade crossing hazard indexes. Hazard indexes are objective formulas for comparing or ranking cross1ngs according to relative hazard or for calculating absolute hazard (conditional expected frequency of grade cross1ng accidents) on an individual crossing basis. Relative and absolute hazard indexes are constructed and compared in performance with some hazard indexes in general use. The DOT-AAR crossing inventory for all public crossings in the United States and the FRA accident data base for 1975 are used. Various measures and displays of performance of hazard indexes in predicting the hazard of crossings as functions of their inventory characteristics and as manifest in the U.S. accident experience of 1975 are given. The levels of performance that may be expected of various hazard indexes in various situations are given. Relative and absolute hazard indexes constructed on this project are exhibited which outperform other hazard indexes tested. Means for shaping a relative hazard index into an absolute hazard index are given. An introductory discussion is provided on the use of accident history in hazard indexes. Preliminary estimates are given of some of the parameters involved in that discussion. Theoretical aspects of this report include some discussions of nonlinear regression and nonlinear discriminant analysis as well as some aspects of empirical Bayesian statistics