Studies of employment-related stress as a risk factor for preterm delivery suggest that contextual factors unrelated to occupation, as well as work-related characteristics, must be examined in assessing this relationship. In this study, the relationship of work and contextual characteristics--assessed at midpregnancy and including scores on an occupational fatigue index--to preterm delivery was examined among 943 black and 425 white low-income multiparous women who were at risk for a poor pregnancy outcome. At 24 to 26 weeks gestational age, a 77-item questionnaire was self-administered to obtain detailed information on sociodemographic and contextual characteristics, home physical activities, and occupational characteristics. Questions in the occupational section of the questionnaire included most of those previously used by Mamelle and coworkers in 1984 and 1987 to construct an occupational fatigue index. The overall preterm delivery rate for black women was 14.0 percent and for white women, 9.6 percent. No relationships were observed between age, education, or marital status and preterm delivery, or between work status, hours per week, transportation, travel time, reliability of child care, or home physical activity and preterm delivery for either black women or white women. Black (but not white) women who continued to work at midpregnancy and who reported being able to take rest breaks when they felt tired had a lower preterm delivery rate (10.4 percent versus 21.9 percent; P = 0.031) compared with those who could or did not. Generally, scores for individual sources and levels of occupational fatigue, as well as total occupational fatigue index scores, were unrelated to preterm delivery in this relatively homogeneous group of low income high-risk women.