A statistical analysis was performed on fatigue test data for aluminum, titanium, steel, and nickel materials. The data for the titanium, steel, and nickel were obtained from spectrum fatigue tests, whereas the data for aluminum were obtained from both constant amplitude and spectrum fatigue tests. The analyzed data consisted of a total of 553 S-N test groups with 2417 specimens and 1288 spectrum test groups with approximately 5000 specimens. The distribution of logarithmic standard deviation of fatigue life for these test groups was analyzed with normal, logarithmic, and two-parameter Weibull probability distribution functions and with 2-deg polynominal equations. The best fit was evaluated using the coefficient of correlation and the chi-squared goodness of fit test. None of the distribution functions or polynomial equations provided the best fit for all of the distributions of the logarithmic standard deviation of fatigue life for the selected sets of test groups. A comparison is also made of three methods of calculating scatter factors.