Eighty-one observational work analyses were conducted to measure stressors independently of worker appraisal in the San Francisco transit system. On the basis of action regulation theory, stress factors were defined as hindrances for task performance due to poor work organization or technological design. Stressors included (a) work barriers, defined as obstacles that cause extra work or unsafe behavior; (b) time pressure; (c) monotonous conditions; and (d) time binding, defined as control over timing. Reliability, measured as interrater agreement, ranged between 80 and 97%, with kappas of .46-.70. Validity analyses were done with 71 transit operators who participated in the observations and 177 operators who were assigned mean line-specific observational stressor measures. High odds ratios (ORs) were found for barriers and psychosomatic complaints (OR = 3.8, p = .00), time pressure and relaxation time needed after work (OR = 3.1, p = .05), and barriers and smoking to cope (OR = 3.8, p = .02). Using observational data in conjunction with self-report data can reduce confounding and improve interpretability of stress and health studies.