A procedure for rating high school or college football teams is developed by applying linear-model methodology to the point spread for each game. The model includes effects for the home-field advantage and for the mean performance levels of the participating teams. The procedure can be modified to use only win-loss information or to ignore victory margins greater than a given margin. When applied to the results of the 1975 college football season, it produced predictions whose accuracy compares favorably with those of sportswriters and bookmakers.