A study examined the types of individuals receiving on-the-job-training and the effect of such training on productivity and earnings. Two years of data from the Michigan Panel Study of Income Dynamics (an 11-year longitudinal study with a sample size of about 200 individuals) concerning the on-the-job-training, labor market experience, and income levels of household heads was used in the study. The final data sample included 1379 observations, with 19 percent of the respondents being female and 91 percent being white. The key variable of the study was the length of training for individual jobs. Data indicated that economic return plays a significant role in the determination of training. Women appear less likely to receive training for a given rate of return to this investment. While the hypothesis of discrimination against women and singles in their access to training cannot be rejected by this test, it appears that discrimination against blacks and the less educated in their access to training is not evident. Further research to refine the measures of the rate of return to training and to explore the mechanisms behind allocation of training is needed. (MN)