Abstract The starting point of this study is the proposition that intensive formation of human capital on the job is the basic proximate reason for the strong degree of worker attachment to the firm in Japan. The greater emphasis on training and retraining, much of it specific to the firm, results also in steeper wage trajectories, due to growth of skills in the firm. Several previous studies viewed the differences between Japanese and U.S. labor markets in the light of the same hypothesis. We explore this insight more thoroughly by a detailed use of microdata for the two countries: we measure wage profiles and turnover in age groups and we test the inverse relation between the two on industry sectors within each of the countries. Numerical estimates of this relation permit us to conclude that as much as two-thirds of the differential in turnover between the two countries is explainable by the differences in the steepness of the profiles. The question remains why the emphasis on human capital formation on the job is so much greater in Japan than in the United States. Our answer is that such emphasis is conditioned by rapid economic growth. More specifically, Japanese labor policies in the firm represent adjustments of worker skills and activities to the very rapid technological changes of the past decades. Using productivity growth indexes for industries in the United States and Japan we test the hypothesis that rapid technical change, which induces greater and continuous training, is in part responsible for steeper profiles and for lesser turnover. The hypothesis is confirmed on the sectoral level in both countries. We conclude that differences in productivity growth between the United States and Japan account for up to 80% of the differences in the steepness of wage profiles, and indirectly for the differences in turnover. Finally, we try to standardize for the cultural background of workers, by observing a sample of Japanese plants in the United States which employ American workers and use Japanese labor policies in recruitment and training. We find that the steeper tenure-wage slopes and lower turnover place this sample closer to Japan than to the United States—about two-thirds of the distance.