For welded ship hull structural members, fatigue strength determined through fatigue test with their scale models and that determined through fatigue test with test specimens of welded joints often are at variance. Accordingly, to find out a clue to more accurate fatigue design of ship hull structural members, the author made a comparative study of fatigue data obtained by the two different specimens or models. During the course of their study, the author (1) established a definition of fatigue crack initiation in scale models, (2) examined how cracks would initiate in certain parts of scale models and affect the behavior of those cracks which later initiate in other parts, (3) determined characteristics of S-N diagram drawn for scale models, (4) compared standard fatigue strength specified for welded joints in steelworks other than ship hull structural members and fatigue strength determined by scale models, (5) examined the manners in which fatigue cracks propagate in scale models and simple shaped test specimens, respectively, and (6) determined the size effect of the scale model on the propagation of fatigue crack.